<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:googleplay="http://www.google.com/schemas/play-podcasts/1.0"><channel><title><![CDATA[JOPRO: Session Notes]]></title><description><![CDATA[Notes, logs, summaries, and other abridged material to aid in asynchronous discourse. For select public and public adjacent sessions, programs, or events. ]]></description><link>https://blog.jopro.org/s/session-notes</link><image><url>https://substackcdn.com/image/fetch/$s_!wQEv!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd0624956-771c-4d90-989a-33440ffe73ed_1080x1080.png</url><title>JOPRO: Session Notes</title><link>https://blog.jopro.org/s/session-notes</link></image><generator>Substack</generator><lastBuildDate>Wed, 18 Nov 2026 03:02:36 GMT</lastBuildDate><atom:link href="https://blog.jopro.org/feed" rel="self" type="application/rss+xml"/><copyright><![CDATA[JOPRO]]></copyright><language><![CDATA[en]]></language><webMaster><![CDATA[jopro@substack.com]]></webMaster><itunes:owner><itunes:email><![CDATA[jopro@substack.com]]></itunes:email><itunes:name><![CDATA[JOPRO]]></itunes:name></itunes:owner><itunes:author><![CDATA[JOPRO]]></itunes:author><googleplay:owner><![CDATA[jopro@substack.com]]></googleplay:owner><googleplay:email><![CDATA[jopro@substack.com]]></googleplay:email><googleplay:author><![CDATA[JOPRO]]></googleplay:author><itunes:block><![CDATA[Yes]]></itunes:block><item><title><![CDATA[Causal world models, and building concepts before building life]]></title><description><![CDATA[Session Notes: Fall 2026 CFRG 07&#8211;08]]></description><link>https://blog.jopro.org/p/cfrg-causal-world-models-and-building</link><guid isPermaLink="false">https://blog.jopro.org/p/cfrg-causal-world-models-and-building</guid><dc:creator><![CDATA[JOPRO]]></dc:creator><pubDate>Tue, 10 Nov 2026 21:53:20 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!5Sy0!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p><em>This post is a rough log for our asynchronous members to view at a glance, rather than an elaborate or high-fidelity capture of the discussion. It covers two sessions.</em></p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!5Sy0!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!5Sy0!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 424w, https://substackcdn.com/image/fetch/$s_!5Sy0!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 848w, https://substackcdn.com/image/fetch/$s_!5Sy0!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 1272w, https://substackcdn.com/image/fetch/$s_!5Sy0!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!5Sy0!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png" width="1254" height="734" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:734,&quot;width&quot;:1254,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:131307,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/224369185?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!5Sy0!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 424w, https://substackcdn.com/image/fetch/$s_!5Sy0!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 848w, https://substackcdn.com/image/fetch/$s_!5Sy0!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 1272w, https://substackcdn.com/image/fetch/$s_!5Sy0!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F221396bc-7a70-41c6-89d4-49b23235bd94_1254x734.png 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image buttonBase-GK1x3M"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image buttonBase-GK1x3M"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>Having finished Stefano Franchi&#8217;s paper on Ashby&#8217;s homeostat and its commentaries ([Notes from sessions 4 and 5 link]), the Cognition Futures Reading Group turned on Thursday to Jonathan Richens and Tom Everitt&#8217;s <a href="https://arxiv.org/abs/2402.10877">&#8220;Robust agents learn causal world models&#8221;</a>, presented at ICLR 2024. Friday was the group&#8217;s first session given over to planning its technical work: what the Friday sessions are for, which tools to use, and a first shared exercise.</p><h2>What was read</h2><p>Richens and Everitt ask whether agents must learn causal models in order to generalize to new domains, or whether other inductive biases are enough. Their answer is a necessity result. An agent that keeps its regret within a bound across a large set of distributional shifts must have learned an approximate causal model of the process that generates its data, and the approximation becomes exact as the agent approaches optimal. The shifts are modeled as local interventions on the environment&#8217;s causal variables. That causal models are sufficient for robust adaptation was already known; the paper argues they are also necessary.</p><p>In the discussion section, the authors revisit the good regulator theorem. It is often read as saying that any good controller must contain a model of the system it controls. The authors point out that the theorem itself only guarantees an optimal regulator that is a deterministic function of the system&#8217;s state, which could be trivial. They present their own result as a sharper, causal version of the theorem. Figure 4 places the result in Pearl&#8217;s causal hierarchy: optimal policies under domain shifts, for a single objective, already cover the interventional level, so they are also enough for shifts in the task.</p><p>The paper is explicit about its limits. The main theorems require robustness to shifts on all of the environment&#8217;s variables, and they cover only decision tasks in which the agent&#8217;s actions do not change the environment. Its discussion also gives AlphaFold as an example of competence without understanding: accurate prediction of protein folding that did not improve understanding of the underlying biochemistry.</p><p>The group read the abstract, the good regulator paragraph, Figure 4 and the conclusion together. Several members found the formal sections hard going, and the group did not take as deep a pass as it did with Franchi.</p><h2>What was discussed</h2><h3>Modeling concepts rather than life</h3><p>Amanda Nelson opened Thursday with what she took from the end of Franchi&#8217;s paper for the group&#8217;s own building work. Modeling life is a grand project, she said, and it would first require deciding what constitutes life. Framing a build as the modeling of a concept, explored openly and with no claim to reproduce a living thing, seemed to her more fruitful.</p><h3>Can a priori structure do technical work?</h3><p>Avery Lim described a conversation with a computer scientist who also studies philosophy. Lim had suggested that his generalized domain primitives might help structure reasoning in language models. The objection was that structures reasoned out in advance can be useful to people as an interface, but cannot capture technical structure, which has to be discovered by statistical exploration. Lim called this the status quo view in computer science. His interlocutor traced it through Wittgenstein and Turing, whom he counts among the roots of cybernetics.</p><p>Jes Parent identified the objection as the bitter lesson in other terms. Nelson asked whether building a model would itself count as enough use of a concept to ground it, in the sense of Wittgenstein&#8217;s language games. Lim said Wittgenstein can be read in more than one way, and proposed tracing the line from Wittgenstein through Turing and logical positivism to see how the divide between concept modeling and object modeling formed, and how it still shapes what researchers treat as viable.</p><h3>Prediction, intervention, and the good regulator</h3><p>Lim drew two threads from the reading. First, he read the paper as separating prediction from robust choice: a good predictor does not need a causal model, but choosing well across a comprehensive range of counterfactual alternatives does. He wondered whether this marks where a priori work earns its place, in organizing a space of causal possibilities and interventions rather than in making specific predictions, which answers part of the objection above. Second, he wanted to look harder at why a causal model is what separates this result from the original framing of the good regulator theorem.</p><p>Lim noted the paper&#8217;s requirement of a comprehensive family of interventions and its AlphaFold example. He also pointed to work, cited in the paper&#8217;s related work, suggesting that trained models can hold internal world models without being given that structure explicitly. He asked what it means for domain and task shifts to overlap with the observational level in Figure 4. Parent read the conclusion aloud: robust decision-making requires a causal model of the data-generating process, which the authors take as evidence that causal world models are a necessary ingredient of general intelligence.</p><h3>Representation and isomorphism</h3><p>Parent mentioned the Brain Inspired group&#8217;s discussion the day before. There, a good part of the conversation turned on how isomorphism is used: whether a given case is an isomorphism, a homomorphism or a functional isomorphism. Janet contributed comments from an engineering perspective, and Kevin Mitchell also took part. Lim added that the discussion raised whether representation is needed for, or implied by, results like this one, a recurring question in embodied cognition and cybernetics. This was raised briefly and not developed.</p><p>Parent also mentioned the AHA bonus seminar with Cathy O&#8217;Neil, which was taking place as the session ended, as something to come back to.</p><h3>What should Fridays be for?</h3><p>Near the end of Thursday, Nelson asked whether the group wants to keep moving between papers or to focus on a particular modeling project. She learns by building, she said, and suggested starting from models that have been discussed for decades, such as the homeostat or one of the most basic models of the neuron. Parent proposed starting with something very simple, then tying it to earlier Orthogonal Research and Education Lab work on developmental Braitenberg vehicles, and then moving into new territory. He asked everyone to spend some time before Friday thinking about a starter task. A perceptron was suggested as a default first target; on Friday, Lim credited the suggestion to Bradly Alicea.</p><p>For the next Thursday reading, Lim offered two papers on Wittgenstein&#8217;s place in the history of AI. Nelson found their AI framing less appealing and preferred Lydia Patton&#8217;s paper on Helmholtz, Wittgenstein and the a priori, which the group chose.</p><h3>Goals for the technical sessions</h3><p>On Friday, Morgan Hough asked what the technical sessions are for: understanding the code yourself, or being able to run simulations. Lim said he wants tools flexible enough for unconventional designs, computationally viable first and user friendly second. Longer term, he wants to model non-trivial adaptation to threat and need, social and environmental, starting from something simple that can be extended. Nelson said she wants to build from simple rules and be able to point to parts of the code and name the concepts they enact, such as a perceptron, a homeostat or feedback. She also wants the behavior being modeled to be describable. Parent described the two as a small, concept-level program and a longer-range one, with the same starting point.</p><h3>Tools and scale</h3><p>Lim had installed NetLogo and suggested following a tutorial together. Nelson leaned toward Python for its wider range of uses. Hough noted that for artificial-life experiments, how far a simulation can run in time is a matter of compute. He pointed to JAX-based packages that let Python code run in parallel on GPUs, and mentioned NEAT, EvoGym and JaxMARL. He described evolutionary robotics, in which an agent&#8217;s morphology and connectivity both change along with its policy, as a large, high-dimensional search space.</p><h3>Related work and the problem of non-trivial models</h3><p>Hough described a computational-life paper his reading group had covered, on the coevolution of sociality and replication. He raised work on modularity in neuroevolution from the mid-2010s, by Jeff Clune and Jean-Baptiste Mouret among others, as relevant to developmental Braitenberg vehicles. He said the work he finds most interesting is Logan Cross&#8217;s on theory of mind in agents: what models agents build of other agents. His own aim is multi-agent models with sociality and theory of mind that are relevant to computational psychiatry, and he noted that such agents also need tasks and organizing principles.</p><p>Lim said that when he has tried toy models, it was hard to get non-trivial behavior: things meant what they did only because he had set them up to, rather than through emergent interaction. Hough called this the balance problem in sociophysics, the difficulty of making statistical-mechanics models interpretable in social terms, and pointed to a 2025 review of the field.</p><h3>A first exercise</h3><p>The group settled on a show and tell for the next Friday. Each person brings a small model they have found or built: a homeostat, a perceptron, a thermostat, a Braitenberg vehicle or something similar. AI-assisted code is fine. Parent asked that the emphasis fall on describing the model: what it does, and whether it achieves what it is supposed to. Lim suggested a shared code repository with a space for each person and a shared notes document, and Parent set up a repository on the JOPRO GitHub for the purpose.</p><h2>Key terms</h2><ul><li><p><strong>Regret bound:</strong> a limit on how much worse an agent&#8217;s policy does than the best available policy.</p></li><li><p><strong>Distributional shift:</strong> in Richens and Everitt, a change to the environment modeled as a local intervention on one or more of its causal variables.</p></li><li><p><strong>Good regulator theorem:</strong> the result, often read as saying that every good regulator of a system must be a model of that system; Richens and Everitt argue its literal content is weaker.</p></li><li><p><strong>Concept simulation:</strong> in Franchi, building a model to explore the consequences of an idea, as distinct from simulating an object.</p></li><li><p><strong>Developmental Braitenberg vehicles:</strong> Braitenberg&#8217;s simple embodied agents, extended so that their nervous systems develop rather than arrive fully formed.</p></li></ul><h2>Open questions carried forward</h2><ul><li><p>Can structure reasoned out in advance do technical work, or only help people interpret? (Lim, Nelson)</p></li><li><p>Does the paper&#8217;s result require representation, or only imply it? (Lim)</p></li><li><p>What does it mean for domain and task shifts to sit where they do in Figure 4? (Lim)</p></li><li><p>How do you get non-trivial behavior from a toy model without building the meaning in? (Lim, Hough)</p></li></ul><h2>Next time</h2><p>On Thursday the group reads Lydia Patton&#8217;s &#8220;Signs, toy models, and the a priori&#8221;; on Friday, the show and tell.</p><div><hr></div><h2>References</h2><h3>This session</h3><ul><li><p>Richens, J. &amp; Everitt, T. (2024). <a href="https://arxiv.org/abs/2402.10877">Robust agents learn causal world models</a>. <em>International Conference on Learning Representations</em> (ICLR 2024).</p></li><li><p>Patton, L. (2009). <a href="https://philarchive.org/archive/PATSTM">Signs, toy models, and the a priori: from Helmholtz to Wittgenstein</a>. <em>Studies in History and Philosophy of Science</em> 40: 281&#8211;289. <a href="https://doi.org/10.1016/j.shpsa.2009.07.004">doi:10.1016/j.shpsa.2009.07.004</a></p></li></ul><h3>Background</h3><ul><li><p>Franchi, S. (2013). Homeostats for the 21st century? Simulating Ashby simulating the brain. <em>Constructivist Foundations</em> 9(1).</p></li><li><p>Dvoretskii, S., Gong, Z., Gupta, A., Parent, J. &amp; Alicea, B. (2020). <a href="https://arxiv.org/abs/2003.07689">Braitenberg Vehicles as Developmental Neurosimulation</a>. arXiv:2003.07689.</p></li></ul><h2>Further &amp; Suggested Reading</h2><p>These readings were not discussed on November 5 or 6. They were gathered after the sessions as starting points for the questions Nelson and Lim raised. The annotations are editorial.</p><h3>Where should a simple first model start?</h3><ul><li><p>Stefan Dvoretskii, Ziyi Gong, Ankit Gupta, Jesse Parent and Bradly Alicea (2020), <a href="https://arxiv.org/abs/2003.07689">&#8220;Braitenberg Vehicles as Developmental Neurosimulation&#8221;</a>. Argues that models which begin as fully formed networks ignore biological development, and proposes Braitenberg vehicles as a minimal embodied system in which a nervous system can develop. It presents three software approaches: a vehicle and genetic algorithm hybrid, multisensory Hebbian learning, and multi-agent models. This is the developmental Braitenberg vehicle work Parent proposed as the bridge from a first simple build.</p></li><li><p>Bradly Alicea, Rishabh Chakrabarty, Stefan Dvoretskii, Ziyi Gong, Akshara Gopi, Avery Lim and Jesse Parent, <a href="https://arxiv.org/abs/2103.05753">&#8220;Continual Developmental Neurosimulation Using Embodied Computational Agents&#8221;</a>. Extends developmental Braitenberg vehicles into a design for learning agents whose development has distinct morphogenetic, critical and learning periods. It proposes growing the network with a genetic algorithm and treats development as a form of continual learning. Relevant to Lim&#8217;s aim of a simple model that can be extended toward non-trivial adaptation.</p></li></ul>]]></content:encoded></item><item><title><![CDATA[Quiescence, preference, and heteronomy: Franchi on Ashby’s homeostat, PT 3]]></title><description><![CDATA[Session Notes: Fall 2026 CFRG 04&#8211;05]]></description><link>https://blog.jopro.org/p/quiescence-preference-and-heteronomy</link><guid isPermaLink="false">https://blog.jopro.org/p/quiescence-preference-and-heteronomy</guid><dc:creator><![CDATA[JOPRO]]></dc:creator><pubDate>Wed, 28 Oct 2026 03:01:02 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!KNgx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p><strong>Week 43 &#183; Sessions 4 and 5 of Fall 2026 &#183; Thursday, October 22 and Friday, October 23, 2026</strong></p><p><em>This post is a rough log for our asynchronous members to view at a glance, rather than an elaborate or high-fidelity capture of the discussion. It covers two sessions. Abridged recordings will be posted to the <a href="https://www.youtube.com/@JOPRO_org">JOPRO YouTube channel</a>.</em></p><p>The Cognition Futures Reading Group continued Stefano Franchi&#8217;s <a href="http://constructivist.info/9/1/093">&#8220;Homeostats for the 21st Century? Simulating Ashby Simulating the Brain&#8221;</a>, published in <em>Constructivist Foundations</em> in 2013. <a href="https://blog.jopro.org/p/the-sleeping-machine-and-the-gradient">Notes from sessions 2 and 3</a> cover Franchi&#8217;s seven theses, the sleeping machine, and Figures 4 and 5. On Thursday the group read on through Franchi&#8217;s case for the homeostat as a concept and his account of heteronomy. On Friday Avery Lim presented work that connects to the reading.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!KNgx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!KNgx!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 424w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 848w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1272w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" width="1456" height="1207" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/cf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1207,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:null,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:null,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:null,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!KNgx!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 424w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 848w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1272w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image buttonBase-GK1x3M"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image buttonBase-GK1x3M"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><h2>What was read</h2><p>On Thursday, Jes Parent recapped sessions 2 and 3, and the group returned to the footnote on Franchi&#8217;s technical and non-technical theses. Thesis 1, general homeostasis, is philosophical; thesis 2 is biological; theses 3 and 4 explain how the first two would be realized. Franchi argues that reading the homeostat as a concept simulation lets attention fall back on theses 1 and 2.</p><p>Franchi then reads Ashby&#8217;s design work, from polystable and multistable systems to the principles of dispersion and localization, as a conceptual exploration of life as passive and contingent. On his account, Ashby is asking what the internal organization of a passive, contingent system would have to be like, and the homeostat serves as a complex thought experiment for answering it.</p><p>From that standpoint, Franchi says the most important of Ashby&#8217;s theses is the first: a complex system running to equilibrium generates complex behavior. Everything else is an explication or a consequence of it. Critical discussion has nonetheless focused on the derived claims, above all ultrastability. Franchi notes that Norbert Wiener, while praising Ashby, reduced homeostasis to a physiological instance of feedback, and that recent work, such as Ezequiel Di Paolo&#8217;s, grafts ultrastability onto neural network controllers. Both, he argues, privilege the object side of Ashby&#8217;s simulation over its conceptual side.</p><p>Franchi suggests the reason is the controversial status of the general homeostasis thesis itself. It insists that life is essentially heteronomous, governed from outside rather than self-governed. That puts it at odds with a long philosophical tradition that separates human from non-human life precisely on the line between autonomous and heteronomous behavior. Footnotes 13 and 14 tie this to a passage from Ashby that Franchi quotes: &#8220;There is yet to be written a book that shall show how all the organism&#8217;s exteriorly directed activities are similarly regulatory.&#8221;</p><p>On Friday the group did not read further. The group will move to the open peer commentaries next.</p><h2>What was discussed</h2><h3>Lim&#8217;s essay and Ashby&#8217;s open book</h3><p>Lim introduced <a href="https://blog.jopro.org/p/the-questions-we-stopped-asking">&#8220;The Questions We Stopped Asking&#8221;</a>, co-written with Parent, the first post in a series on what Lim calls mediated gestalt inference. Lim described it as a historical overview of a holistic, whole-system kind of inquiry that has been devalued, and said the next post will take up &#8220;domain primitives.&#8221; Lim connected the essay to Ashby&#8217;s claim that even high-order human drives, such as playing chess, can be described as homeostatic, and to the &#8220;yet to be written&#8221; book quoted above. Parent compared the essay&#8217;s project to the way the 1943 paper by Rosenblueth, Wiener and Bigelow reclaimed the term teleology, and stressed that it is not an argument against empiricism or the scientific method.</p><h3>Simulating objects and simulating concepts</h3><p>Lim asked whether Franchi&#8217;s distinction means that applying Ashby naively, at the level of cells, is wrong, while applying him at higher levels of organization is more accurate. Parent read Franchi differently: the distinction is about not reducing the homeostat to a brain simulation, because Ashby was also working through philosophical questions. Amanda Nelson pointed to Franchi&#8217;s remark that technical work in the cognitive sciences is sometimes philosophy in disguise. She described concept simulation as exploratory, playing with abstract concepts rather than aiming only at accurate reproduction, and noted Franchi&#8217;s point that a good model does both. Lim extended this: simulating a concept may improve understanding of the object itself. Nelson thought that was consistent with Franchi, and said she wished he had gone further on what the ideal combination looks like.</p><h3>The sleeping machine and the dark room problem</h3><p>Nelson asked how closely the sleeping machine relates to predictive processing&#8217;s dark room problem: if an organism&#8217;s aim were to minimize prediction error, it could simply shut itself in a dark, predictable room. That looks anti-life, and she noted it is presumably an old objection with many replies. Lim asked whether anyone has related it to sensory hypersensitivity, as in autism or migraine, where a predictable environment is sought. Nelson said organisms sometimes do face something like the dark room, but that it is generally taken as a sign something has gone wrong. Parent connected it to differing degrees of sensory permeability and to how many variables a person can hold at once.</p><h3>Quiescence or set point?</h3><p>Lim asked why Ashby and Franchi frame quiescence as the goal, rather than maintaining a set point with rest as the end of a cycle. Parent restated the question as whether a system reaches a stable local equilibrium in a larger landscape rather than the whole system going still, and noted that the word &#8220;quiescence&#8221; appears only once in Franchi&#8217;s article. Parent proposed collecting the group&#8217;s questions before reading the commentaries, so they can be compared with what the commentators and Franchi&#8217;s response say.</p><h3>Heteronomy and what counts as life</h3><p>Parent read the heteronomy passage and asked Nelson what it meant to her. Nelson connected it to her earlier point that a drive toward sleep looks anti-life. For Maturana and Varela, autonomy, a system&#8217;s self-production and self-regulation, is what defines life, and many philosophers have described something like a life force. If life is heteronomous in Ashby&#8217;s sense, she asked, how do we distinguish life from non-life? Parent said he wants to look into why the tradition settled on autonomy as the dividing line, as a review of the literature.</p><h3>Preference as homeostasis</h3><p>Lim suggested the group was circling preference and set points within subsystems, and proposed that preference and autonomy may themselves be the maintenance of a set point, dependent on homeostatic regulation. It is hard to explain why someone plays chess, Lim said, if the drive is only to do the minimum and return to rest. Lim pointed to individual differences in drives, such as the Big Five, need for cognition, and need for affect, and to a Frontiers paper on why people enjoy horror films, and suggested homeostasis may describe complex behavior better when it concerns information rather than survival. Parent placed this under what he calls the non-uniformity of phenomenological experience. Lim proposed grounding the next discussion in a triad of appetite for diversity, for intensity, and for quiescence, and planned to go back to what Ashby himself wrote about quiescence. This was raised as a direction to develop and was not settled.</p><h3>The quiescence trap</h3><p>On Friday, Parent shared a new interoception paper by Micah Allen and colleagues that names a &#8220;quiescence trap&#8221;: measuring interoception at rest, typically through a single channel such as heartbeat detection, when the field&#8217;s interest is in how bodily signals shape emotional life. Nelson read the term as an ecological validity problem, and Morgan Hough noted the paper&#8217;s use of video of people at great heights to induce arousal. The word is used differently than in Franchi, but the overlap was noted.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!oYDL!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!oYDL!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 424w, https://substackcdn.com/image/fetch/$s_!oYDL!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 848w, https://substackcdn.com/image/fetch/$s_!oYDL!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 1272w, https://substackcdn.com/image/fetch/$s_!oYDL!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 1456w" sizes="auto, 100vw"><img src="https://substackcdn.com/image/fetch/$s_!oYDL!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png" width="1456" height="777" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/ec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:777,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:585191,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/222433543?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!oYDL!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 424w, https://substackcdn.com/image/fetch/$s_!oYDL!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 848w, https://substackcdn.com/image/fetch/$s_!oYDL!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 1272w, https://substackcdn.com/image/fetch/$s_!oYDL!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fec542ea1-28df-4cdc-bdb4-ca4bc8a8d1d5_1884x1006.png 1456w" sizes="auto, 100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container restack-image buttonBase-GK1x3M"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button tabindex="0" type="button" class="pencraft pc-reset pencraft icon-container view-image buttonBase-GK1x3M"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><h3>Lim&#8217;s presentation</h3><p>Lim presented on universal domain primitives and how they relate to Ashby, to Franchi&#8217;s reading, and to earlier work on allostatic kinds and allostasis machines.  </p><p>Nelson asked whether the project is a kind of counter-Enlightenment argument, and whether a rationalist, modeling approach has limits for questions about life and consciousness. Lim answered that the argument is for integration rather than a rejection of Enlightenment methods, pointing to Hegel&#8217;s idea of sublation, and allowed that the essay&#8217;s title overstates the case. Bradly Alicea said the work partly matches the lab&#8217;s existing research, needs more connecting to it, and then needs testing.</p><h2>Key terms</h2><ul><li><p><strong>Heteronomy:</strong> behavior governed by something outside the system; the counterpart of autonomy.</p></li><li><p><strong>General homeostasis thesis:</strong> Ashby&#8217;s first thesis, that a complex system running to equilibrium generates complex behavior.</p></li><li><p><strong>Concept simulation:</strong> in Franchi, building a model to explore the consequences of an idea, as distinct from simulating an object such as the brain.</p></li><li><p><strong>Dark room problem:</strong> the objection that an agent minimizing prediction error should seek a dark, unchanging room.</p></li><li><p><strong>Quiescence trap:</strong> Allen and colleagues&#8217; term for studying brain-body interaction only at rest.</p></li><li><p><strong>Mediated gestalt inference:</strong> Lim and Parent&#8217;s name for holistic inquiry into connections that run through intermediaries.</p></li></ul><h2>Open questions carried forward</h2><ul><li><p>Why frame the homeostat&#8217;s goal as quiescence rather than maintaining a set point? (Lim)</p></li><li><p>If life is heteronomous in Ashby&#8217;s sense, what distinguishes living from non-living systems? (Nelson)</p></li><li><p>Can preference itself be modeled as set-point maintenance? (Lim)</p></li><li><p>What would the ideal combination of object and concept simulation look like? (Nelson)</p></li><li><p>What are the limits of a modeling approach to questions about life and consciousness? (Nelson)</p></li></ul><h2>Next time</h2><p>The group moves to the open peer commentaries and Franchi&#8217;s response.</p><p class="button-wrapper" data-attrs="{&quot;url&quot;:&quot;https://blog.jopro.org/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe now&quot;,&quot;action&quot;:null,&quot;class&quot;:null}" data-component-name="ButtonCreateButton"><a class="button primary" href="https://blog.jopro.org/subscribe?"><span>Subscribe now</span></a></p><div><hr></div><h2>References</h2><h3>This session</h3><ul><li><p>Franchi, S. (2013). Homeostats for the 21st century? Simulating Ashby simulating the brain. <em>Constructivist Foundations</em> 9(1): 93&#8211;101. Open peer commentaries and the author&#8217;s response follow in the same issue. <a href="http://constructivist.info/9/1/093">http://constructivist.info/9/1/093</a></p></li><li><p>Lim, A. &amp; Parent, J. (2026). <a href="https://blog.jopro.org/p/the-questions-we-stopped-asking">The Questions We Stopped Asking</a>. JOPRO, October 19, 2026.</p></li><li><p>Rosenblueth, A., Wiener, N. &amp; Bigelow, J. (1943). Behavior, purpose and teleology. <em>Philosophy of Science</em> 10(1): 18&#8211;24.</p></li><li><p>Maturana, H. R. &amp; Varela, F. J. (1980). <em>Autopoiesis and Cognition: The Realization of the Living</em>. Reidel.</p></li><li><p>Allen, M., et al. The quiescence trap of measuring interoception at rest.</p></li></ul><h3>Earlier notes</h3><ul><li><p><a href="https://blog.jopro.org/p/the-sleeping-machine-and-the-gradient">The sleeping machine and the gradient: Franchi on Ashby&#8217;s homeostat, PT 2</a>. Its further reading covers Andy Clark on the dark room problem and Peter Sterling on allostasis.</p></li></ul><h2>Further &amp; Suggested Reading</h2><p>These readings were not discussed on October 22 or 23. They were gathered after the sessions as starting points for the questions Lim and Nelson raised. The annotations are editorial.</p><h3>What does a method set aside in order to work?</h3><ul><li><p>Jes Parent (2026), <a href="https://blog.jesparent.com/p/declared-abstraction-on-bracketing">&#8220;Declared Abstraction: On Bracketing and Saying What You Leave Out&#8221;</a>. A second look at the 1943 Rosenblueth, Wiener and Bigelow paper, arguing that its authors explicitly set aside the internal organization of the systems they studied and said what that cost. Relevant to Franchi&#8217;s claim that Ashby&#8217;s readers kept his technical theses and quietly set aside the philosophical ones, and to Nelson&#8217;s question about where modeling stops.</p></li></ul><h3>Where do the right primitives come from?</h3><ul><li><p>Battleday and Gershman&#8217;s paper on the easy and hard problems of AI for science, discussed and linked in <a href="https://blog.jopro.org/p/the-questions-we-stopped-asking">&#8220;The Questions We Stopped Asking&#8221;</a>. They argue that formulating a problem, including choosing its primitives, is the hard part of science. Lim cited it as a starting point for the domain-primitives work.</p></li></ul>]]></content:encoded></item><item><title><![CDATA[The sleeping machine and the gradient: Franchi on Ashby's homeostat, PT 2 ]]></title><description><![CDATA[Session Notes: Fall 2026 CFRG 02-03]]></description><link>https://blog.jopro.org/p/the-sleeping-machine-and-the-gradient</link><guid isPermaLink="false">https://blog.jopro.org/p/the-sleeping-machine-and-the-gradient</guid><dc:creator><![CDATA[JOPRO]]></dc:creator><pubDate>Fri, 16 Oct 2026 20:44:04 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!KNgx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p><em>This post is a rough log for our asynchronous members to view at a glance, rather than an elaborate or high-fidelity capture of the discussion. It covers two sessions. Abridged recordings will be posted to the</em> <a href="https://www.youtube.com/@JOPRO_org">JOPRO YouTube channel</a>.</p><div><hr></div><p>The Cognition Futures Reading Group continued Stefano Franchi&#8217;s &#8220;Homeostats for the 21st Century? Simulating Ashby Simulating the Brain,&#8221; published in <em>Constructivist Foundations</em> in 2013. <a href="https://blog.jopro.org/p/simulating-objects-simulating-concepts">Notes from session 1</a> cover the paper&#8217;s opening sections and its distinction between simulating objects and simulating concepts. On Thursday the group considered a first build for the Friday sessions and two recent talks on AI and human relationships. On Friday it returned to the paper.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!KNgx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!KNgx!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 424w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 848w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1272w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" width="1456" height="1207" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/cf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1207,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:459214,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/218387480?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!KNgx!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 424w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 848w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1272w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a><figcaption class="image-caption">Franchi&#8217;s abstract</figcaption></figure></div><h2>What was read</h2><p>On Friday, Parent recapped the first session for anyone new, and the group read on from the end of &#8220;Simulating objects vs. simulating concepts&#8221; to paragraph 10 of &#8220;The homeostat as a concept,&#8221; then turned to Figures 4 and 5 from the paper&#8217;s later sections.</p><p>Franchi notes that most readers treat Ashby&#8217;s homeostat as a proxy object, a drastically simplified brain, and argues it should also be read as a proxy concept. He reduces Ashby&#8217;s theory to seven theses. The first four: a complex system running to equilibrium generates complex behavior; a system is in equilibrium when it keeps its essential variables at stable levels; it reaches equilibrium through ultrastability, rearranging the connections between its parts when those variables become unstable; and it achieves ultrastability through a stochastic step function that randomly steps through the parameters connecting its parts. From his experiments Ashby derived three more: a fully connected system takes an exceedingly long time to reach equilibrium, so a real system needs weak connections among its parts, and its subsystems will typically communicate through the environment rather than directly.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!Hr4j!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!Hr4j!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 424w, https://substackcdn.com/image/fetch/$s_!Hr4j!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 848w, https://substackcdn.com/image/fetch/$s_!Hr4j!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 1272w, https://substackcdn.com/image/fetch/$s_!Hr4j!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!Hr4j!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png" width="1456" height="979" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:979,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:450540,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/220762558?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!Hr4j!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 424w, https://substackcdn.com/image/fetch/$s_!Hr4j!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 848w, https://substackcdn.com/image/fetch/$s_!Hr4j!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 1272w, https://substackcdn.com/image/fetch/$s_!Hr4j!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2c4cee9a-4539-47b6-b863-bee9ef8facda_1538x1034.png 1456w" sizes="100vw"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><p>Franchi calls the first thesis, generalized homeostasis, Ashby&#8217;s most original and least warranted idea, and the second his least original but most solid. The third and fourth are Ashby&#8217;s core technical contribution. In a footnote he explains that readers interested in building artifacts or explaining biology have focused on theses 3 and 4 and let 1 and 2 slide, and that reading the homeostat as a concept simulation brings the philosophical and biological theses back into view.</p><p>The passage closes with William Grey Walter&#8217;s nickname for the homeostat, <em>machina sopora</em>, the sleeping machine. Franchi finds the jibe superficial but telling: on Ashby&#8217;s view, the homeostat &#8220;will act as much as is needed in order to go back to sleep.&#8221; It is passive, because its activity is a by-product of seeking rest, and contingent, because its search for equilibrium runs through random processes.</p><p>Figure 4 shows the minimal two-unit homeostat as an agent and an environment, the agent negatively self-connected and the environment not, joined by sensory and motor transducers. Figure 5 places that setup inside a Braitenberg-style vehicle with one sensor and one wheel, moving in an environmental gradient between upper and lower bounds. Franchi maps his three experimental outcomes onto it: the vehicle rests at the optimum, oscillates back and forth between the bounds, or settles into constant motion that keeps its essential variable within tolerance without reaching the top.</p><h2>What was discussed</h2><h3>A first build for the Friday sessions</h3><p>On Thursday, Amanda Nelson proposed using Franchi&#8217;s paper as a guide for a first build: replicating the homeostat and its small experiments, and possibly the vehicle in Figure 5. Her reasoning was that a setup someone else has already specified would show the group what it does not yet know about building one. There was further discussion of Franchi&#8217;s footnote on Braitenberg&#8217;s vehicles, whose simplest types are routinely used to introduce reactive architectures that wire a sensor directly to a motor, and noted that Franchi&#8217;s robot replaces that direct link with a homeostatic unit. The proposal was welcomed as a candidate; nothing has been scheduled. Nelson also said she is keen to read the peer commentaries.</p><h3>What quiescence can explain</h3><p>On Friday, Lim asked how far Ashby&#8217;s model reaches. Some human drivers relate to allostatic adaptation, and others are aesthetic play and social games, such as watching horror films. These might be modeled as reaching a set point and then returning to rest, but it is not clear how quiescence as the goal accounts for them. Lim also suggested that ultrastability might be reached through several entry and exit points, different perspectives on the same system, and that this could supply the alternatives a system needs before it can evaluate goals at all. Parent connected this to Lim&#8217;s earlier question, &#8220;where do goals come from?&#8221;, raised when the group read the telic states paper, and to the idea that information can only be transmitted as a choice among alternatives. These were raised as ideas to pick up and were not developed further.</p><h3>Franchi&#8217;s Figures 4 and 5</h3><div class="image-gallery-embed" data-attrs="{&quot;gallery&quot;:{&quot;images&quot;:[{&quot;type&quot;:&quot;image/png&quot;,&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/98962721-e74f-4716-ac55-2575e1a004b2_468x1166.png&quot;},{&quot;type&quot;:&quot;image/png&quot;,&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/01879a7b-a2e7-4d31-b6c5-af2321614184_464x474.png&quot;}],&quot;caption&quot;:&quot;Figures (click to expand)&quot;,&quot;alt&quot;:&quot;&quot;,&quot;staticGalleryImage&quot;:{&quot;type&quot;:&quot;image/png&quot;,&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/c3c79471-1a5f-4d33-986d-c3144d5068a2_1456x720.png&quot;}},&quot;isEditorNode&quot;:true}"></div><p>Bradly Alicea walked the group through the figures. In Figure 4 he traced the double feedback: one loop between agent and environment through the sensory and motor transducers, and a second through the agent&#8217;s feedback onto itself. He read Figure 5 as gradient following: as the vehicle climbs the gradient the signal strengthens and reinforces its path, so a simple vehicle can lock onto a local maximum, such as light reflecting off a bright surface, instead of the actual source. Lim summarized this as local reward versus global modeling. Alicea suggested richer feedback could let a vehicle distinguish sources, compared the simple case to being in an information bubble, and said the figure raises more questions than answers.</p><h2>Related Discussion</h2><h3>Moral reasoning and self-report</h3><p>Parent recapped a talk by Cody Turner at a Boston University  event, based on Turner and Nir Eisikovits&#8217;s paper &#8220;<a href="https://link.springer.com/article/10.1007/s43681-026-01007-4">Programmed to Please</a>,&#8221; which treats AI sycophancy as an Aristotelian vice with moral and epistemic harms. He shared notes from a Data x Direction members also attended, on how AI sycophancy differs from the human kind and how training on human preferences rewards affirming answers.</p><p>Nelson asked whether the question of a model&#8217;s moral reasoning is already answered by the reasoning some models display, or by the instructions they are given. Parent answered with Julia Haas and colleagues&#8217; roadmap for evaluating moral competence in large language models and its facsimile problem: a model may produce appropriate moral outputs without reasoning from morally relevant considerations. He also considered the matter as a self-report problem, familiar from clinical psychology, where what someone says about their own state may be unreliable or misleading on its own.</p><p>Parent added that discussions of AI ethics outside technical fields often treat a model as a fixed entity, and that this misses how models work: there is no fixed persona, context is limited and gets compacted, and memory is far from perfect. He tied this to Sacco, Sakthivadivel, and Levin on topological limits to self-organization, which the group read earlier this year, and floated the idea of a teaching resource on how LLMs are structured, written for non-specialists.</p><h3>Language as a social relation</h3><p>Hough argued that chatbots, which have no goals or desires of their own, nonetheless pull on evolved needs for social relation. He pointed to Emily Bender&#8217;s work on pragmatics and to the psychiatrist Matthew Nour, who he noted is now at Microsoft Research. He also asked what linguists are saying about the effect of LLMs on human language, suggesting lingbuzz as a place to look.</p><h3>Turkle&#8217;s <em>Artificial Intimacy</em></h3><p>Parent attended an event for Sherry Turkle&#8217;s new book, <em>Artificial Intimacy: Who We Become When We Talk to Machines</em>, and shared three points: her account of a companion-chatbot founder crediting Turkle&#8217;s earlier work as the basis of their business; her repeated question of what the end game is, which Parent connected to his own framing of destinations; and her stated decision in this book to center humans deliberately.</p><div class="pullquote"><p>Technology is seductive when its affordances meet our human vulnerabilities - Sherry Turkle, <a href="https://www.techhuman.org/resource-bank/bzx8eoajuethhu9uz7cax1jomx77em">Alone Together</a></p></div><h3>LLMs as a cognitive virus</h3><p>Parent flagged a recent <a href="https://arxiv.org/abs/2609.03344">preprint</a> that models the spread of LLM use through a population with an epidemiological analogy, including tipping points and &#8220;cognitive immunization.&#8221; Hough connected it to multi-compartment diffusion models and asked to discuss it when there is time.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!J_BK!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!J_BK!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 424w, https://substackcdn.com/image/fetch/$s_!J_BK!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 848w, https://substackcdn.com/image/fetch/$s_!J_BK!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 1272w, https://substackcdn.com/image/fetch/$s_!J_BK!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 1456w" sizes="auto, 100vw"><img src="https://substackcdn.com/image/fetch/$s_!J_BK!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png" width="520" height="414.27710843373495" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/b713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1058,&quot;width&quot;:1328,&quot;resizeWidth&quot;:520,&quot;bytes&quot;:278136,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/220762558?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!J_BK!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 424w, https://substackcdn.com/image/fetch/$s_!J_BK!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 848w, https://substackcdn.com/image/fetch/$s_!J_BK!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 1272w, https://substackcdn.com/image/fetch/$s_!J_BK!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fb713b27a-ebcf-4b87-a7eb-6d12d309bb58_1328x1058.png 1456w" sizes="auto, 100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><div><hr></div><h2>Key terms</h2><ul><li><p><strong>Homeostat:</strong> Ashby&#8217;s electro-mechanical device of four fully connected units, each able to randomly reset its connection weights.</p></li><li><p><strong>Essential variables:</strong> variables a system must keep within bounds to remain viable.</p></li><li><p><strong>Ultrastability:</strong> reaching equilibrium by rearranging internal connections whenever essential variables go out of bounds.</p></li><li><p><strong>Uniselector:</strong> Ashby&#8217;s name for the random step mechanism that does the rearranging.</p></li><li><p><em><strong>Machina sopora</strong></em><strong>:</strong> Grey Walter&#8217;s &#8220;sleeping machine&#8221; nickname for the homeostat.</p></li><li><p><strong>Heteronomy:</strong> behavior governed by something outside the system; the counterpart of autonomy.</p></li><li><p><strong>Braitenberg Type 1 vehicle:</strong> a simple robot with one sensor wired to one motor.</p></li><li><p><strong>Facsimile problem:</strong> Haas and colleagues&#8217; term for a model imitating reasoning without genuine understanding.</p></li></ul><p class="button-wrapper" data-attrs="{&quot;url&quot;:&quot;https://blog.jopro.org/p/the-sleeping-machine-and-the-gradient/comments&quot;,&quot;text&quot;:&quot;Leave a comment&quot;,&quot;action&quot;:null,&quot;class&quot;:null}" data-component-name="ButtonCreateButton"><a class="button primary" href="https://blog.jopro.org/p/the-sleeping-machine-and-the-gradient/comments"><span>Leave a comment</span></a></p><h2>Open questions carried forward</h2><ul><li><p>Could Franchi&#8217;s homeostat and vehicle serve as a Friday build, and what would the group need to learn first? (Nelson)</p></li><li><p>How far can a quiescence-seeking model account for play, aesthetics, and social drives? (Lim)</p></li><li><p>Where do goals come from in an Ashbian system? (Lim)</p></li><li><p>What kind of feedback would let a simple vehicle tell a reflection from a source? (Alicea)</p></li><li><p>What are linguists saying about how LLMs are changing human language? (Hough)</p></li><li><p>Can a model&#8217;s moral reasoning be assessed beyond what it reports about itself? (Nelson, Parent)</p></li></ul><h2>Next time</h2><p>The group continues Franchi&#8217;s paper. Whether it finishes the target article or moves to the open peer commentaries first is still open.</p><div><hr></div><div class="subscription-widget-wrap-editor" data-attrs="{&quot;url&quot;:&quot;https://blog.jopro.org/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe&quot;,&quot;language&quot;:&quot;en&quot;}" data-component-name="SubscribeWidgetToDOM"><div class="subscription-widget show-subscribe"><div class="preamble"><p class="cta-caption">Thank you for reading - - Subscribe for free to support JOPRO&#8217;s work and stay informed of our latest updates.</p></div><form class="subscription-widget-subscribe"><input type="email" class="email-input" name="email" placeholder="Type your email&#8230;" tabindex="-1"><input type="submit" class="button primary" value="Subscribe"><div class="fake-input-wrapper"><div class="fake-input"></div><div class="fake-button"></div></div></form></div></div><div><hr></div><h2>References</h2><h3>This session</h3><ul><li><p>Franchi, S. (2013). Homeostats for the 21st century? Simulating Ashby simulating the brain. <em>Constructivist Foundations</em> 9(1): 93&#8211;101. Open peer commentaries and the author&#8217;s response follow in the same issue. <a href="http://constructivist.info/9/1/093">http://constructivist.info/9/1/093</a></p></li><li><p>Braitenberg, V. (1984). <em>Vehicles: Experiments in Synthetic Psychology</em>. MIT Press. <a href="https://mitpressbookstore.mit.edu/book/9780262521123">https://mitpressbookstore.mit.edu/book/9780262521123</a></p></li><li><p>Turner, C. &amp; Eisikovits, N. (2026). Programmed to please: The moral and epistemic harms of AI sycophancy. <em>AI and Ethics</em> 6. <a href="https://link.springer.com/article/10.1007/s43681-026-01007-4">https://link.springer.com/article/10.1007/s43681-026-01007-4</a></p></li><li><p>Haas, J., Bridgers, S., Manzini, A., et al. (2026). A roadmap for evaluating moral competence in large language models. <em>Nature</em> 650: 565&#8211;573. <a href="https://doi.org/10.1038/s41586-025-10021-1">https://doi.org/10.1038/s41586-025-10021-1</a></p></li><li><p>Sacco, F., Sakthivadivel, D. &amp; Levin, M. (2026). Topological constraints on self-organization in locally interacting systems. <em>Philosophical Transactions of the Royal Society A</em> 384(2320): 20250011. <a href="https://doi.org/10.1098/rsta.2025.0011">https://doi.org/10.1098/rsta.2025.0011</a></p></li><li><p>Amir, N., Tiomkin, S. &amp; Langdon, A. (2026). Goals and the structure of experience. <em>Philosophical Transactions of the Royal Society A</em> 384(2320): 20250004.</p></li><li><p>Turkle, S. (2026). <em>Artificial Intimacy: Who We Become When We Talk to Machines</em>. Little, Brown. <a href="https://www.hachettebookgroup.com/titles/sherry-turkle/artificial-intimacy/9780316573962/">https://www.hachettebookgroup.com/titles/sherry-turkle/artificial-intimacy/9780316573962/</a></p></li><li><p>Sol&#233;, R., et al. (2026). Large-language models as a cognitive virus. arXiv:2609.03344. <a href="https://arxiv.org/abs/2609.03344">https://arxiv.org/abs/2609.03344</a></p></li></ul><h2>Further &amp; Suggested Reading</h2><p>These readings were not discussed on October 15 or 16. They were gathered after the sessions as starting points for the questions Nelson, Lim, and Alicea raised. The annotations are editorial.</p><h3>Can a model&#8217;s account of its own reasoning be trusted?</h3><ul><li><p>Miles Turpin, Julian Michael, Ethan Perez, and Samuel Bowman (2023), <a href="https://arxiv.org/abs/2305.04388">&#8220;Language Models Don&#8217;t Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting&#8221;</a>. The authors bias models toward particular answers through features of the prompt, such as reordering multiple-choice options, and find that the step-by-step explanations rationalize those answers without mentioning the bias. A direct test of Nelson&#8217;s question about whether displayed reasoning shows what drives an output.</p></li><li><p>Anthropic (2025), <a href="https://www.anthropic.com/research/reasoning-models-dont-say-think">&#8220;Reasoning models don&#8217;t always say what they think&#8221;</a>. A follow-up that tests the faithfulness of reasoning models, which show their working by design, and reports largely negative results: models often used hints in reaching an answer without mentioning them. The Haas et al. roadmap makes a related point, noting it is unclear whether reasoning traces reflect the processes that actually produce a response.</p></li></ul><h3>Can a system that seeks rest account for activity?</h3><ul><li><p>Andy Clark (2013), <a href="https://doi.org/10.1017/S0140525X12000477">&#8220;Whatever next? Predictive brains, situated agents, and the future of cognitive science&#8221;</a>. <em>Behavioral and Brain Sciences</em> 36(3): 181&#8211;204. Clark&#8217;s target article on the brain as a prediction machine. In his commentary on Franchi, Tom Froese cites it for the &#8220;dark room&#8221; problem, the question of why an agent minimizing surprise would do anything at all, as a modern version of the sleeping-machine worry. Relevant to Lim&#8217;s question about horror films and play.</p></li><li><p>Peter Sterling (2004), &#8220;Principles of allostasis: Optimal design, predictive regulation, pathophysiology and rational therapeutics,&#8221; in J. Schulkin (ed.), <em>Allostasis, Homeostasis, and the Costs of Adaptation</em>, Cambridge University Press. Allostasis treats regulation as predictive rather than fixed around a set point. Robert Lowe&#8217;s commentary on Franchi draws on it to argue that organisms must sometimes operate outside their comfort zone, which bears on Lim&#8217;s mention of allostatic adaptation.</p></li></ul><h3>On the vehicle</h3><ul><li><p>Ezequiel Di Paolo (2000), &#8220;Homeostatic adaptation to inversion of the visual field and other sensorimotor disruptions,&#8221; in <em>From Animals to Animats 6</em>, MIT Press: 440&#8211;449. Franchi cites this as the leading example of grafting Ashby&#8217;s ultrastability onto a neural network controller. A worked example of the kind of build Nelson proposed.</p></li></ul><h3>From the article&#8217;s own commentaries</h3><p>Several of the open peer commentaries, which the group has not yet reached, take up these questions directly. Matthew Egbert and Inman Harvey both argue that keeping essential variables within bounds is not passive, using a pole balanced on a fingertip and a homeostatically controlled bicycle. Takashi Ikegami asks where the &#8220;dice&#8221; behind Ashby&#8217;s random rewiring come from and whether they are biased, which sits close to Lim&#8217;s entry-points idea. Franchi&#8217;s response sets out seven senses of &#8220;passivity&#8221; and answers the bicycle and pole examples.</p><div><hr></div><p></p>]]></content:encoded></item><item><title><![CDATA[Simulating objects, simulating concepts: Franchi on Ashby’s homeostat, PT 1 | CFRG 26Fall01]]></title><description><![CDATA[Session Notes: Fall 2026 CFRG 01]]></description><link>https://blog.jopro.org/p/simulating-objects-simulating-concepts</link><guid isPermaLink="false">https://blog.jopro.org/p/simulating-objects-simulating-concepts</guid><dc:creator><![CDATA[JOPRO]]></dc:creator><pubDate>Thu, 01 Oct 2026 19:43:56 GMT</pubDate><enclosure url="https://substackcdn.com/image/fetch/$s_!KNgx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p><em>This post is a rough log for our asynchronous members to view at-a-glance, rather than an elaborate or high-fidelity capturing of the discussion. An abridged video recording of the session will be made public.</em></p><div><hr></div><p>On October 1, 2026, the Cognition Futures Reading Group began reading Stefano Franchi&#8217;s &#8220;<a href="https://www.researchgate.net/publication/277596924_Homeostats_for_the_21st_Century_Simulating_Ashby_Simulating_the_Brain">Homeostats for the 21st Century</a>? Simulating Ashby Simulating the Brain,&#8221; published in <em>Constructivist Foundations</em> in 2013. The article appeared as a target article with seven open peer commentaries (from Inman Harvey, Helge Malmgren, Robert Lowe, Tom Froese, Matthew Egbert, David Vernon, and Takashi Ikegami) and a response from Franchi. Amanda Nelson, <span class="mention-wrap" data-attrs="{&quot;name&quot;:&quot;Bradly Alicea&quot;,&quot;id&quot;:20563261,&quot;type&quot;:&quot;user&quot;,&quot;url&quot;:null,&quot;photo_url&quot;:&quot;https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fbucketeer-e05bbc84-baa3-437e-9518-adb32be77984.s3.amazonaws.com%2Fpublic%2Fimages%2Ff6bacd2c-e67c-46c6-b751-6f652a814108_236x187.png&quot;,&quot;uuid&quot;:&quot;ac39dcf9-39d1-43f8-b397-4ccec2765d17&quot;}" data-component-name="MentionToDOM"></span>, Morgan Hough and <span class="mention-wrap" data-attrs="{&quot;name&quot;:&quot;Jes Parent&quot;,&quot;id&quot;:28022075,&quot;type&quot;:&quot;user&quot;,&quot;url&quot;:null,&quot;photo_url&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/2b9944de-52ad-4321-80e4-7d2800cc8533_689x689.png&quot;,&quot;uuid&quot;:&quot;eaadd8f7-49a6-4fb8-b457-0cab4e340f9b&quot;}" data-component-name="MentionToDOM"></span> attended.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!KNgx!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!KNgx!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 424w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 848w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1272w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1456w" sizes="100vw"><img src="https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png" width="1456" height="1207" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/cf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:1207,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:459214,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:false,&quot;topImage&quot;:true,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/218387480?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!KNgx!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 424w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 848w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1272w, https://substackcdn.com/image/fetch/$s_!KNgx!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fcf776dc1-6095-41ea-a8ac-fb5237bd20c4_1496x1240.png 1456w" sizes="100vw" fetchpriority="high"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a><figcaption class="image-caption">Franchi&#8217;s abstract</figcaption></figure></div><p>Nelson proposed the paper. She had been looking at classic cybernetics models, early neuron models and W. Ross Ashby&#8217;s homeostat among them, and found in Franchi a paper about what such models are for.</p><p>This was the first meeting in this series for the Fall 2026 term. </p><h2>What was read</h2><p>The group worked slowly through the abstract and the opening sections, stopping towards the end of the section titled &#8220;Simulating objects vs. simulating concepts.&#8221;</p><p>Franchi&#8217;s abstract sets up the argument. Ashby&#8217;s technical results, such as ultrastability, have been taken up in recent cognitive science, but his broader thesis that homeostatic adaptation governs all aspects of all life has been left behind. Franchi argues that this thesis makes life fundamentally heteronomous, and that it conflicts both with the autopoiesis framework used by Ashby&#8217;s recent defenders and with the Western tradition&#8217;s emphasis on autonomy. A definition of heteronomy was shared in the chat: action governed by a force outside the individual, the counterpart of autonomy. Parent noted how the paper cuts into that tradition&#8217;s treatment of autonomy as what defines life.</p><p>The opening sections draw the distinction the session centered on. In <em>simulating objects</em>, a model is judged by how closely its behavior matches the real thing. Franchi&#8217;s example is Newell and Simon&#8217;s early AI work, which compared machine performance on tasks like chess with human performance on the same tasks. In <em>simulating concepts</em>, the model is &#8220;a tool to think with&#8221; (Franchi&#8217;s phrase, after Papert&#8217;s Logo microworlds) and closeness to the real world is not the measure. Franchi holds that most successful projects use both, but that the two should be kept apart analytically, because the focus on object behavior can hide the philosophical work underneath. He puts it bluntly: technical work in the cognitive sciences is sometimes &#8220;philosophy in disguise.&#8221; The section closes by applying this to Ashby: the homeostat is usually read as a simplified proxy for the brain, and Franchi argues it can only be appreciated fully if it is also read as a proxy concept.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!eBnp!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!eBnp!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 424w, https://substackcdn.com/image/fetch/$s_!eBnp!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 848w, https://substackcdn.com/image/fetch/$s_!eBnp!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 1272w, https://substackcdn.com/image/fetch/$s_!eBnp!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 1456w" sizes="auto, 100vw"><img src="https://substackcdn.com/image/fetch/$s_!eBnp!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png" width="1456" height="976" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:976,&quot;width&quot;:1456,&quot;resizeWidth&quot;:null,&quot;bytes&quot;:1599854,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/218387480?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!eBnp!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 424w, https://substackcdn.com/image/fetch/$s_!eBnp!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 848w, https://substackcdn.com/image/fetch/$s_!eBnp!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 1272w, https://substackcdn.com/image/fetch/$s_!eBnp!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F189052f0-f2e2-4829-8445-cf65596f931b_1608x1078.png 1456w" sizes="auto, 100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><div><hr></div><h2>What was discussed</h2><h3>Concept simulation, and what it is good for</h3><p>Nelson opened with an example of her own that the paper does not use: Randall Beer&#8217;s work using cellular automata to demonstrate a concept like precariousness. A model of that kind does not need to resemble anything in particular in the world. She also noted Franchi&#8217;s point that the most useful models combine the two approaches, and that in cognitive science a model should be grounded in biology in some fashion.</p><p>Later, Nelson said she valued Franchi&#8217;s account of philosophy as the study of &#8220;being as such,&#8221; as opposed to a loose synonym for thinking. Read that way, she said, philosophy does not have to happen only in one&#8217;s head or on paper, and a computer can be a tool for it.</p><p>Parent noted an appreciation of the word &#8220;simulation&#8221; being used in a particular fashion, as distinct from &#8220;model&#8221; or &#8220;idea.&#8221; He asked what non-brain object simulations look like, offering hurricanes and weather, stock prices, and a heart condition as candidates. He also connected the Newell and Simon comparison to his interest in substrate agnosticism: two substrates attempt the same task, and the match between them is taken as evidence for the theory. He asked whether a concept could be carried across substrates, then noted the question may not fit, since concept simulations are not judged by that kind of validity. In an aside, he observed that Franchi&#8217;s 2013 remark that identifying the mind with symbol-manipulating software had &#8220;long gone out of fashion&#8221; reads differently in the current AI moment.</p><div class="captioned-image-container"><figure><a class="image-link image2 is-viewable-img" target="_blank" href="https://substackcdn.com/image/fetch/$s_!s1jN!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png" data-component-name="Image2ToDOM"><div class="image2-inset"><picture><source type="image/webp" srcset="https://substackcdn.com/image/fetch/$s_!s1jN!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 424w, https://substackcdn.com/image/fetch/$s_!s1jN!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 848w, https://substackcdn.com/image/fetch/$s_!s1jN!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 1272w, https://substackcdn.com/image/fetch/$s_!s1jN!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 1456w" sizes="auto, 100vw"><img src="https://substackcdn.com/image/fetch/$s_!s1jN!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png" width="488" height="410.3636363636364" data-attrs="{&quot;src&quot;:&quot;https://substack-post-media.s3.amazonaws.com/public/images/3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png&quot;,&quot;srcNoWatermark&quot;:null,&quot;fullscreen&quot;:null,&quot;imageSize&quot;:null,&quot;height&quot;:888,&quot;width&quot;:1056,&quot;resizeWidth&quot;:488,&quot;bytes&quot;:641039,&quot;alt&quot;:null,&quot;title&quot;:null,&quot;type&quot;:&quot;image/png&quot;,&quot;href&quot;:null,&quot;belowTheFold&quot;:true,&quot;topImage&quot;:false,&quot;internalRedirect&quot;:&quot;https://blog.jopro.org/i/218387480?img=https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png&quot;,&quot;isProcessing&quot;:false,&quot;align&quot;:null,&quot;offset&quot;:false}" class="sizing-normal" alt="" srcset="https://substackcdn.com/image/fetch/$s_!s1jN!,w_424,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 424w, https://substackcdn.com/image/fetch/$s_!s1jN!,w_848,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 848w, https://substackcdn.com/image/fetch/$s_!s1jN!,w_1272,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 1272w, https://substackcdn.com/image/fetch/$s_!s1jN!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3a8caf4d-ac87-493f-a970-48fbed93fe6f_1056x888.png 1456w" sizes="auto, 100vw" loading="lazy"></picture><div class="image-link-expand"><div class="pencraft pc-display-flex pc-gap-8 pc-reset"><button class="pencraft pc-reset pencraft icon-container restack-image buttonBase-tqxFUX" tabindex="0" type="button"><svg aria-hidden="true" width="20" height="20" viewBox="0 0 20 20" fill="none" stroke-width="1.5" stroke="var(--color-fg-primary)" stroke-linecap="round" stroke-linejoin="round" xmlns="http://www.w3.org/2000/svg" class="icon-noB79L"><g><path d="M2.53001 7.81595C3.49179 4.73911 6.43281 2.5 9.91173 2.5C13.1684 2.5 15.9537 4.46214 17.0852 7.23684L17.6179 8.67647M17.6179 8.67647L18.5002 4.26471M17.6179 8.67647L13.6473 6.91176M17.4995 12.1841C16.5378 15.2609 13.5967 17.5 10.1178 17.5C6.86118 17.5 4.07589 15.5379 2.94432 12.7632L2.41165 11.3235M2.41165 11.3235L1.5293 15.7353M2.41165 11.3235L6.38224 13.0882"></path></g></svg></button><button class="pencraft pc-reset pencraft icon-container view-image buttonBase-tqxFUX" tabindex="0" type="button"><svg xmlns="http://www.w3.org/2000/svg" width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" class="lucide lucide-maximize2 lucide-maximize-2 icon-noB79L"><polyline points="15 3 21 3 21 9"></polyline><polyline points="9 21 3 21 3 15"></polyline><line x1="21" x2="14" y1="3" y2="10"></line><line x1="3" x2="10" y1="21" y2="14"></line></svg></button></div></div></div></a></figure></div><h3>Ultrastability and machine learning</h3><p>Alicea raised a connection between Ashby&#8217;s mechanism and machine learning. Ashby&#8217;s homeostat adapts by randomly switching its connection parameters when a unit&#8217;s essential variable goes out of bounds, a stochastic step function. Alicea compared this to step-like functions in neural networks, such as the Heaviside or ReLU function, acting on a weight matrix, and described ultrastability as close to the way a neural network is pruned. He framed this as part of an answer to the question of whatever happened to cybernetics: some of it went on to motivate machine learning. He suggested there is probably a paper to be written on these connections. This was floated as a line worth pursuing; the group did not examine the comparison in detail.</p><h3>Is the distinction new, and how far does it go?</h3><p>Parent asked whether Franchi&#8217;s object/concept distinction is standard in computational modeling or an unusual move. Alicea said it echoes a familiar criticism of modeling: a model is not the thing modeled. He thought Franchi motivates the distinction more than necessary, since the homeostat is plainly an analogy for an organism rather than a replica of one.</p><p>Alicea then gave the distinction a more specific stake. In cognitive science, he said, a model often represents two things at once: something like brain or network activity, and the constructs that activity is supposed to produce, such as attention or memory. Researchers switch between the two as if nothing needed to sit between them, and whether the construct has been captured correctly often goes unasked.</p><p>Nelson agreed the distinction is correct and that each approach has its own benefits and should be analyzed separately. But she argued there is more to say than the paper does: a model has to be fruitful in both domains to be truly fruitful, and good experimentation keeps both in view. In her words, &#8220;the whole is greater than the sum of its parts.&#8221; She did not recall what Franchi says about the homeostat on this point specifically.</p><p>Parent built on Nelson&#8217;s point in closing. He suggested thinking in degrees: how far a given simulation deliberately moves away from a concrete object comparison, such as checking a machine&#8217;s output against a human&#8217;s, and toward pure concept, and whether it later comes back. On this view neither end is better. What matters is being intentional about the distance. He wondered whether this could become a framework or even a metric. This was raised as an idea to pick up, and was not developed further in the session.</p><h2>Open questions carried forward</h2><ul><li><p>Is the object/concept distinction a standard one in computational modeling, or is Franchi drawing it in an unusual way? Alicea offered one answer; Parent&#8217;s question was left open.</p></li><li><p>What does Franchi say about the homeostat specifically on the need to work in both modes? Nelson raised this and the reading had not yet reached it.</p></li><li><p>Could Parent&#8217;s degrees-of-abstraction idea be made concrete?</p></li><li><p>Does Alicea&#8217;s comparison between ultrastability and machine learning hold up in detail?</p></li></ul><p>Next time, the group will continue the paper with the section &#8220;The homeostat as a concept,&#8221; where Franchi sets out his reading of Ashby&#8217;s theses, ahead of the simulation results, the commentaries, and the author&#8217;s response.</p><p class="button-wrapper" data-attrs="{&quot;url&quot;:&quot;https://blog.jopro.org/subscribe?&quot;,&quot;text&quot;:&quot;Subscribe now&quot;,&quot;action&quot;:null,&quot;class&quot;:null}" data-component-name="ButtonCreateButton"><a class="button primary" href="https://blog.jopro.org/subscribe?"><span>Subscribe now</span></a></p><div><hr></div><h2>CFRG Context: Previous Discussions </h2><p><em>This section is editorial context drawn from the group&#8217;s earlier sessions this year. None of it was discussed on October 1.</em></p><ul><li><p><strong>Dynamical cognitive science.</strong> Earlier this year the group read Raja&#8217;s introduction to a <em>Topics in Cognitive Science</em> issue on dynamical cognitive science, followed by Randall Beer&#8217;s &#8220;On the Proper Treatment of Dynamics in Cognitive Science&#8221; from the same issue. Nelson&#8217;s October 1 example comes from a different line of Beer&#8217;s work, his cellular automaton models of autopoiesis and precariousness.</p></li><li><p><strong>World models.</strong> Several readings came from the <a href="https://royalsocietypublishing.org/rsta/issue/384/2320">2026 </a><em><a href="https://royalsocietypublishing.org/rsta/issue/384/2320">Philosophical Transactions of the Royal Society A</a></em><a href="https://royalsocietypublishing.org/rsta/issue/384/2320"> </a>theme issue on world models in natural and artificial intelligence. That issue includes a paper by four members of the group (Alicea, Hough, Nelson, and Parent) recasting the <a href="https://royalsocietypublishing.org/rsta/article/384/2320/20250007/481683/A-good-regulator-may-provide-a-world-model-for">Every Good Regulator Theorem as a framework for world modeling</a>. The theorem is Conant and Ashby&#8217;s, so the group has been working with Ashby&#8217;s ideas before this paper. Other readings from the issue included Sacco, Sakthivadivel, and Levin on topological limits to self-organization in locally interacting systems, including autoregressive language models; Amir, Tiomkin, and Langdon on goal-relative &#8220;telic states&#8221;; Hofstadter on whether there is an &#8220;I&#8221; in AI; and Battleday and Gershman on easy and hard problems in AI for science.</p></li><li><p><strong>The agent as unit of explanation.</strong> Several of those readings, along with a paper by Tom Froese on methodological individualism, questioned whether a single, central agent is the right unit for explaining purposeful behavior. Franchi&#8217;s heteronomy thesis approaches autonomy from another direction. Froese is also one of the commentators on Franchi&#8217;s article, which the group has yet to reach.</p></li></ul><h2>References</h2><p><strong>This session</strong></p><ul><li><p>Franchi, S. (2013). Homeostats for the 21st century? Simulating Ashby simulating the brain. <em>Constructivist Foundations</em> 9(1): 93&#8211;101. Open peer commentaries and the author&#8217;s response follow in the same issue (pp. 102&#8211;124). https://constructivist.info/9/1/093</p></li><li><p>Beer, R. D. &amp; Di Paolo, E. A. (2023). The theoretical foundations of enaction: Precariousness. <em>Biosystems</em>. https://www.sciencedirect.com/science/article/abs/pii/S0303264722002040 </p></li><li><p>Ashby, W. R. (1960). <em>Design for a brain</em>. Second edition. John Wiley &amp; Sons, New York. (The source of the homeostat as Franchi discusses it.)</p></li></ul><p><strong>Background</strong></p><ul><li><p>Alicea, B., Hough, M., Nelson, A. &amp; Parent, J. (2026). A &#8216;good&#8217; regulator may provide a world model for intelligent systems. <em>Philosophical Transactions of the Royal Society A</em> 384(2320): 20250007. https://doi.org/10.1098/rsta.2025.0007</p></li><li><p>Beer, R. D. (2023). On the proper treatment of dynamics in cognitive science. <em>Topics in Cognitive Science</em>. https://doi.org/10.1111/tops.12686 </p></li><li><p>Sacco, F., Sakthivadivel, D. &amp; Levin, M. (2026). Topological constraints on self-organization in locally interacting systems. <em>Philosophical Transactions of the Royal Society A</em>. https://doi.org/10.1098/rsta.2025.0011 </p></li><li><p>Amir, N., Tiomkin, S. &amp; Langdon, A. (2026). Goals and the structure of experience. <em>Philosophical Transactions of the Royal Society A</em> 384(2320): 20250004. </p></li><li><p>Hofstadter, D. (2026). Is there an &#8216;I&#8217; in AI? <em>Philosophical Transactions of the Royal Society A</em> 384(2320): 20240527. https://doi.org/10.1098/rsta.2024.0527 &#8224;</p></li><li><p>Battleday, R. M. &amp; Gershman, S. J. (2026). Artificial intelligence for science: The easy and hard problems. <em>Philosophical Transactions of the Royal Society A</em> 384(2320): 20240530. https://doi.org/10.1098/rsta.2024.0530 </p></li><li><p>Favela, L.H. and Raja, V. (2026), Dynamical Cognitive Science! Wherefore Art Thou?. Top. Cogn. Sci., 18: e70042. <a href="https://doi.org/10.1111/tops.70042">https://doi.org/10.1111/tops.70042</a></p><p></p></li></ul><p class="button-wrapper" data-attrs="{&quot;url&quot;:&quot;https://blog.jopro.org/p/simulating-objects-simulating-concepts/comments&quot;,&quot;text&quot;:&quot;Leave a comment&quot;,&quot;action&quot;:null,&quot;class&quot;:null}" data-component-name="ButtonCreateButton"><a class="button primary" href="https://blog.jopro.org/p/simulating-objects-simulating-concepts/comments"><span>Leave a comment</span></a></p><div><hr></div><h2>Further &amp; Suggested Reading</h2><p><em>These readings were not discussed on October 1. They were gathered following the session as starting points for the questions Parent, Nelson, and Alicea raised about simulating objects versus simulating concepts. Several of the article&#8217;s own commentaries take up similar questions. The annotations are editorial.</em></p><h3>Mapping the space between object and concept</h3><p><strong>Barbara Webb (2001), <a href="https://doi.org/10.1017/S0140525X01000127">&#8220;Can robots make good models of biological behaviour?&#8221;</a></strong> <em>Behavioral and Brain Sciences</em> 24(6). Webb separates simulation models, which represent a hypothesis about a target system, from other activities also called modeling. She then sets out seven dimensions on which such models vary: biological relevance, level of organization, generality, abstraction, structural accuracy, behavioral match, and physical medium. She argues that no position in that space is the single correct one, and that a good methodology states where a model sits and why. This is the closest existing match to Parent&#8217;s call for intentionality about how far a simulation moves from its object.</p><p><strong>Michael Weisberg (2013), </strong><em><strong><a href="https://academic.oup.com/book/3141/chapter-abstract/143990987">Simulation and Similarity</a></strong></em><strong><a href="https://academic.oup.com/book/3141/chapter-abstract/143990987">, chapter 7, &#8220;Modeling Without a Specific Target&#8221;</a>.</strong> Weisberg argues that building a single model to study a specific target is only one part of modeling practice. This chapter covers generalized, hypothetical, and targetless modeling, the last being the study of a model with no target at all. His examples include cellular automata and the Game of Life, which places targetless modeling near the kind of example Nelson raised.</p><p><strong>Arturo Rosenblueth and Norbert Wiener (1945), <a href="https://doi.org/10.1086/286874">&#8220;The Role of Models in Science&#8221;</a>.</strong> <em>Philosophy of Science</em> 12(4). The cybernetic starting point for this debate. The authors hold that no substantial part of the universe can be grasped without abstraction, which they define as replacing the part under study with a simpler model of similar structure, and they divide models into formal and material kinds. They also supply a limit case for the object end of any spectrum: &#8220;the best material model for a cat is another, or preferably the same cat.&#8221;</p><p><strong>Richard Levins (1966), <a href="https://www.semanticscholar.org/paper/The-strategy-of-model-building-in-population-Levins/ee0659d2e67584f03a3af01809cfa8aae79f9ada">&#8220;The Strategy of Model Building in Population Biology&#8221;</a>.</strong> <em>American Scientist</em> 54(4). Levins argues that models trade generality, realism, and precision against one another, and describes three strategies that each sacrifice one of the three. He also holds that a satisfactory theory is usually a cluster of models. That point bears on Nelson&#8217;s argument that good modeling has to work in more than one mode, since Levins locates strength in a family of models rather than in any single one.</p><h3>Simulation as thought experiment</h3><p><strong>Ezequiel Di Paolo, Jason Noble, and Seth Bullock (2000), <a href="https://www.semanticscholar.org/paper/Simulation-models-as-opaque-thought-experiments-Paolo-Noble/60cd6ebb92ff5b22a13d65554a5dc566eeb8f2be">&#8220;Simulation models as opaque thought experiments&#8221;</a>.</strong> <em>Artificial Life VII</em>. The authors set two views against each other: simulations as computational thought experiments, and simulations as realistic simulacra. They propose that simulations are opaque thought experiments, in which the consequences follow from the premises in a way that is not obvious and has to be revealed through systematic enquiry. This offers one answer to why concept work needs a computer at all, which connects to Nelson&#8217;s point about computers as tools for philosophy.</p><p><strong>Andrew Pickering (2010), </strong><em><strong>The Cybernetic Brain: Sketches of Another Future</strong></em><strong>.</strong> Pickering reads the homeostat as &#8220;ontological theatre,&#8221; a device that stages an ontology, and argues that read this way the homeostat is a contribution to philosophy. It is a historian&#8217;s route to Franchi&#8217;s claim that technical work can be philosophy in disguise, and the book is cited by several of the commentators. A <a href="https://eusp.org/sites/default/files/archive/centres/respub/Pickering_article.pdf">short essay version</a> is available, and <em>Constructivist Foundations</em> ran a <a href="https://constructivist.info/13/3/398.roberts">review of the book</a>.</p><p><strong>Randall Beer (2003), <a href="https://doi.org/10.1177/1059712303114001">&#8220;The Dynamics of Active Categorical Perception in an Evolved Model Agent&#8221;</a>.</strong> <em>Adaptive Behavior</em>. Alongside a detailed analysis of a simple evolved agent, Beer argues in a section titled &#8220;Frictionless Brains&#8221; that cognitive science needs simpler idealized models of complete brain, body, and environment systems, in the way physics uses frictionless planes. He states that the aim of such models is conceptual understanding rather than empirical prediction. This is Beer&#8217;s own account of what concept models are for, from the author of Nelson&#8217;s example. His later <a href="https://doi.org/10.1177/1059712320931595">&#8220;Some historical context for minimal cognition&#8221;</a> (2021) traces how the approach developed.</p><p><strong>Mark Bedau (1999), &#8220;Can unrealistic computer models illuminate theoretical biology?&#8221;</strong> <em>Proceedings of the 1999 Genetic and Evolutionary Computation Conference Workshop Program</em>. A short paper that asks the question in its title directly, and is often cited alongside Di Paolo, Noble, and Bullock. No stable online copy was found.</p><p><strong>Christopher Langton (1989), <a href="https://www.routledge.com/Artificial-Life-Proceedings-Of-An-Interdisciplinary-Workshop-On-The-Synthesis/Langton/p/book/9780367152772">&#8220;Artificial Life&#8221;</a>.</strong> The opening chapter of the first Artificial Life proceedings. Langton framed the field as placing life as we know it within the larger space of life as it could be. It is the programmatic statement behind much of the concept-side simulation work in artificial life.</p><p><strong>Seymour Papert (1980), </strong><em><strong><a href="https://mindstorms.media.mit.edu/">Mindstorms: Children, Computers, and Powerful Ideas</a></strong></em><strong>.</strong> The source of the &#8220;tool to think with&#8221; idea Franchi draws on. Papert describes the Logo turtle as an &#8220;object-to-think-with.&#8221; The book is freely available online through the MIT Media Lab.</p><h3>How the two modes relate</h3><p><strong>Barbara Webb (2009), <a href="https://doi.org/10.1177/1059712309339867">&#8220;Animals Versus Animats: Or Why Not Model the Real Iguana?&#8221;</a></strong> <em>Adaptive Behavior</em>. Webb distinguishes models of specific animal systems from the exploration of invented artificial animals, and asks how simulating animals that do not exist can teach us about real biology. Her answer is that animat research, where it bears on biology, should be treated as model building like any other. It is a direct test of Nelson&#8217;s claim that a good model has to work in both modes.</p><p><strong>Evelyn Fox Keller (2000), <a href="https://doi.org/10.1086/392810">&#8220;Models of and models for: Theory and practice in contemporary biology&#8221;</a>.</strong> <em>Philosophy of Science</em> 67 (Supplement). Keller describes models of gene pathways as models <em>of</em> mechanisms and, at the same time, models <em>for</em> generating new questions and experiments. The pairing gives a name to Nelson&#8217;s two modes as two directions a single model can face.</p><p><strong>Mary Morgan and Margaret Morrison, eds. (1999), </strong><em><strong><a href="https://philpapers.org/rec/MORMAM-7">Models as Mediators: Perspectives on Natural and Social Science</a></strong></em><strong>.</strong> The editors argue that models mediate between theory and world while remaining partly independent of both. This suggests a third way to place the homeostat, alongside object and concept.</p><p><strong>William Wimsatt (1987), &#8220;False Models as Means to Truer Theories.&#8221;</strong> In Nitecki and Hoffman, eds., <em>Neutral Models in Biology</em>, Oxford University Press. Wimsatt argues that idealized and strictly false models do real work in theory building, partly by showing where and how they fail. It is reprinted in his <em>Re-Engineering Philosophy for Limited Beings</em> (2007).</p><p><strong>Howard Pattee (1989), <a href="https://doi.org/10.4324/9780429032769-3">&#8220;Simulations, Realizations, and Theories of Life&#8221;</a>.</strong> In the same Artificial Life proceedings as Langton. Pattee distinguishes computer simulations of life from computer realizations of it, and asks how either relates to a theory that separates the living from the non-living. This bears on Parent&#8217;s substrate question: at what point does running a concept on a new substrate stop being a simulation of it?</p><h3>Models and the constructs they claim to capture</h3><p><strong>Olivia Guest and Andrea Martin (2021), <a href="https://doi.org/10.1177/1745691620970585">&#8220;How Computational Modeling Can Force Theory Building in Psychological Science&#8221;</a>.</strong> <em>Perspectives on Psychological Science</em> 16(4). The authors argue that building a computational model forces researchers to analyze, specify, and formalize intuitions that would otherwise go unexamined, which they call open theory. They describe inference as a path that must step through theory, specification, and implementation. This is relevant to Alicea&#8217;s point about researchers switching between a model&#8217;s network activity and the constructs it is said to produce.</p><p><strong>Iris van Rooij and Giosu&#232; Baggio (2021), <a href="https://doi.org/10.1177/1745691620970604">&#8220;Theory Before the Test: How to Build High-Verisimilitude Explanatory Theories in Psychological Science&#8221;</a>.</strong> <em>Perspectives on Psychological Science</em> 16(4). The authors argue that a focus on experimental effects can lose sight of the psychological capacities those effects are meant to explain. They revisit Marr&#8217;s levels of analysis, the classic framework for the level-switching Alicea described, and show how theoretical analysis can establish a theory&#8217;s plausibility before it meets data.</p><h3>From Franchi&#8217;s own footnotes</h3><p><strong>Peter Asaro (2011), <a href="https://www.researchgate.net/publication/266408841_Computers_as_Models_of_the_Mind_On_Simulations_Brains_and_the_Design_of_Early_Computers">&#8220;Computers as Models of the Mind: On Simulations, Brains and the Design of Early Computers&#8221;</a>.</strong> In Franchi and Bianchini, eds., <em>The Search for a Theory of Cognition</em>. Franchi cites this for a more detailed analysis of the uses of computer simulation. A related <a href="https://doi.org/10.25365/oezg-2008-19-4-4">open-access article in German</a> (2008) discusses correspondence in which Alan Turing urged Ashby to simulate the homeostat on the ACE computer rather than build a dedicated machine, an early precedent for Franchi&#8217;s virtual homeostat.</p><p><strong>Antoni Gomila and Vincent M&#252;ller (2012), <a href="https://arxiv.org/abs/2505.20339">&#8220;Challenges for artificial cognitive systems&#8221;</a>.</strong> <em>Journal of Cognitive Science</em> 13(4). Franchi cites this for the theoretical basis behind object simulations.</p>]]></content:encoded></item></channel></rss>