Sessions for NeuroAI Minor - ░ ░ ░ ░ ░

### NeuroAI minor ## Philosophy delves **Lecturer**: Sonia de Jager ##### (Erasmus School of Philosophy, PhD in philosophy of NLP) Email: [email protected] **Times**: Friday afternoons.  **Format**: presentation and class activity, followed by discussion.  --- ![[Sessions for NeuroAI Minor bio1.png]] --- ![[Sessions for NeuroAI Minor uva IAS llms.png]] ![[Sessions for NeuroAI Minor uva ias question llms.png]] --- The words “actually” and “merely” are doing a lot of lifting here. “Actually understand” perhaps suggests we already have a sense of what understanding entails, which I suspect has to do with how we perceive human agents as capable of determining the contextual relevance and applicability of epistemic structures to reality-tracking. If so: (L)LMs (from w2v to transformers) are the ultimate context-machines, “they” understand, or are representations of understanding (like a dynamic library), because we have built them precisely for that. If something else: Q = what is “understanding”? --- “Merely” suggests the manipulation of statistical patterns does not entail understanding. However, in the context of AIF, the argument is that this is all living, surviving systems do: track the patterns they are able to distinguish by sense perception by way of compressions that, for their purposes, respect symmetries (invariances) accurate enough to ensure perceptual continuity, planning, survival, etc. So, Q = relevance-compression and limits of physics/mathematics as representational systems and for what purposes. --- ![[Sessions for NeuroAI Minor bio6.png]] --- ![[Sessions for NeuroAI Minor bio2.png]] --- ![[Sessions for NeuroAI Minor bio3.png]] --- ![[Sessions for NeuroAI Minor bio4.png]] --- ![[Sessions for NeuroAI Minor bio5.png]] --- --- --- --- ###### Sessions will follow the course concepts and structure. No reading required... #### ... but: do come prepared to these sessions with questions! ![[1000 Images website/aperiodic tiling hat.png]] Aperiodic tiling “hat”, David Smith, Joseph Myers, Chaim Goodman-Strauss, 2023. Many readings recommended, you can follow the links on these slides on my website (www.n-o.ooo). --- ##### Examples of class concepts that we will dive into ![[1000 Images website/Mario image highlights.png]] --- ### Sessions + Week 1 (after onboarding): **NeuroAI** + What is intelligence? + What are explanations? (versus proofs, predictions, etc.) + Week 2: **Neurons & Models** + What is a (good) model? How do models gain traction on reality? + Week 3: **Neuroanatomy & Neurodynamics** + What is computation? + Do brains compute? --- + Week 4: **Encoding & Decoding** + Sense-making and meaning + How systems couple to one another + Week 5: **Learning & Plasticity** + Learning and metalearning; levels of abstraction + Evolution of learning (Baldwin effect) + Week 6: **Structure &/vs Function** + Patterns: e.g., symmetry as invariance + Function as invariance & invariances of function! --- + Week 7: **Integration week** + Story so far + Open questions + Story onwards + Week 8: **Review week** + Conceptual loose ends + AOB? + Week 9: **Release week** + What is consciousness? + Can machines be conscious? --- + Week 10: **Arrival** + End + Course feedback + Try to think ahead: what would you have liked to have learned, if you imagine yourself 10 weeks forward, having traveled all these incursions? + As Mario points out: your feedback and participation will be invaluable throughout! --- ### Week 1: **Neuroscience + “AI”** ![[Healing_Grid, Ryota Kanai, Utrecht, 2005.jpg]] ##### *Healing Grid*, Ryota Kanai, Utrecht, 2005. %% note: why healing grid, patterns, physiology. Also note: induced motion perception in NNs, Watanabe et al. w/ rotating snakes. %% --- ##### Why philosophy? Because when we reach the outer edges of scientific knowledge, all that is left are philosophical (mathematical)$^1$ questions. And, main question for much of cog-neuro + AI = ### what is intelligence? ##### $^1$ These two are highly interrelated and conceptually difficult to differentiate, remind me to return to this if you are interested in hearing more. Relates to: [[All learning is learning language]]. --- ##### Asking _what_ (and why) is often unsuitable for approaching complex concepts, because they are context-dependent, subject to many interacting constraints, etc. Our intuition for “what” seems to be based on the fact that we’re creatures with pretty prominent object-permanence proclivities, we like to point to things, and have things pointed out.$^1$ But where to _point_ to when we ask what intelligence is? It’s a dynamic, multifaceted phenomenon... and there are many instantiations, approaches, etc. #### _How_ is often more hypothesis-friendly and scientific. ##### $^1$ You see? I am (at)tempted to answer a “what/why” question here. --- It’s similar to asking what maths is, and: ##### Maths, like recipes, has both ingredients and method ... we can’t understand [it] unless we talk about the _way it is done_. [Just like with intelligence] ... It’s a bit of a chicken-and-egg question [because] maths is defined by the techniques [i.e., the *how*] it uses to study things [i.e., the *what*], and that the things it studies are determined by those techniques. Eugenia Cheng, _Cakes, Custard and Category Theory_, 2015, p. 10. --- #### Likewise, intelligence can be understood as the various ways (the _how_) intelligent creatures display/perform/plan/etc. behavior (the _what_), which they recognize$^1$ as intelligent because because they themselves possess (and evolve) that capacity. ##### $^1$ Please pay close attention to the word *re-cognize* here: it is not only a sign of recursivity in the mind, but particularly: in distributed, dialogical (intelligent) creatures! --- ### A philosopher will also tell you that most scientists are _realists_. What is realism? --- ##### Literature phil. of neuroscience, “haptic realism”: ![[1000 Images website/chirimuuta MIT press.png|600]] --- ### Chirimuuta: ## Simplification through analogy (e.g., brain = computer) is useful, but... + Analogy obfuscates everything that is _not_ the same about two different things. + More on analogy: _Surfaces and Essences_ by Hofstadter & Sander (2013), Basic Books. --- ![[Sessions for NeuroAI Minor electron micrograph.png]] --- ##### Literature phil. of AI (title echoing Dennett), posthumanism: ![[1000 Images website/Screenshot 2026-07-06 at 18.25.26.png|500]] ##### Hayles, N. Katherine. _Bacteria to AI: Human Futures with Our Nonhuman Symbionts_. University of Chicago Press, 2025. --- ![[1000 Images website/hayles anthropocentrism.png|600]] --- ##### Whether or not we agree, food for thought: ![[1000 Images website/hayles cognition consciousness.png|600]] ##### Hayles, pp. 2-4. --- ##### Literature, conceptual overview phil. of sci., (excellent for thinking what/why/how questions in dynamics/causality): ![[1000 Images website/alicia context book.png|500]] --- ![[1000 Images website/juarrero constraints.png|700]] ##### Juarrero, p. 87. --- ![[1000 Images website/juarrero recursion.png|600]] ##### Juarrero, p. 88. --- ##### E.g., BZ reaction: ![[1000 Images website/juarrero bz reaction.png|700]] ##### Juarrero, p. 91. --- The BZ chemical reaction is a great example because as Juarrero notes: “more so than physical convection cells, hyperloop-creating constraints revive the specter of self-cause, the [bane] of Western philosophy and science.” [[Autopoiesis]]. “The two autocatalytic cycles in the BZ reaction intertwine to self-produce a set of mereological constraints. The hypercycle as a whole stabilizes and governs its individual reactions (including the locally autocatalytic fourth step) which, when combined, enable the hypercyle in the first place. This recursive and Escher-like “strange loop” (Hofstadter 1979) makes BZ hyper-cycles exemplars of Kant’s self-organizing self-cause—“a form of causality unknown to us” (Juarrero-Roque 1985).” (p. 91). --- As Mario (who also looks at the BZ reaction) points out: “The meaning of the mechanism analogy in the study of organisms and brain function has to be put under adequate light to be serviceable.” (v, preface to _Invariants of behavior_, 2009). What constraint-focused rather than local/building-block-focused approaches can give us is attention to what adequate light is: the word “gol” is not an energy-carrying phenomenon per se, but can be understood to have tremendous potential energy in the right context (a footbal stadium, for example). --- ### Behind stunning patterns in nature there are invariant properties of components and invariant rules, at once defiant and elusive. Once revealed, an invariant transmutes into a mathematical crystal. Negrello 2009, pp. 63-4. (See also: Turing on morphogenesis, or: why Fibonacci in many phenomena? “Lazy” natural processes? Where do gradients come from?!) --- [[Operational coherence]] as the realist+pragmatist road. ### Questions of scale; history, etc. lurk in the background: ![[1000 Images website/hasok chang.png|200]] ##### Chang, Hasok: _Inventing Temperature: Measurement and Scientific Progress_ (Oxford Studies in Philosophy of Science). Oxford University Press, USA, Oxford University Press USA, Oxford, 2004. --- ![[1000 Images website/hasok chang boiling water.png]] ##### Chang, p. 8. --- Or think about how Turing had to be there for the computer to become what we know it to be, today. If Church had published the lambda calculus without anyone taking _mechanical_ notice, things might have been very different today (notes in [[2026-07-28]]). But if none of these examples convince you why philosophy needs science and science needs philosophy... --- ##### ... just to give to very popular (science) zeitgeist instances: Sean Carroll or, for example, Led Fridman. ##### Both latter and former used to dismiss philosophy as a side distraction only to have turned into self-made philosophers and proponents of philosophical inquiry themselves! --- Or, about scale... Geoffrey West of [[Allometric scaling]] fame: invariances such as scales and laws are easy. What is difficult is the speculative “and if so, what if...?” E.g., what if we discover (new/other/unimaginable) limits to intelligence? Or limits to sociality? [[Dialogue between great-grandmother and great-grandson]] Example. --- ##### Again, main question for much of cog-neuro + AI = ### what is intelligence? ##### (almost no wrong answers here)  (And even though nobody has defined what intelligence exactly means, all manner of researchers are using the term to refer to, e.g., developments in AI control, interpretability, futurology, etc.) --- #### Is it... Whatever leads to survival (and reproduction)?  Plastic adaptation to novelty?  (Relevantly contextualized) compression?  Prediction (which succeeds over other predictions)?  Simplification?  + ... what is missing? --- #### Is it... Distributed (across groups, generations, etc.) or highly local (only brain? brain and body?)?  Something that itself keeps changing/evolving? (+ Is there such a thing as “progress”/linearity?)  “In the eye of the beholder”?  Even a useful concept?  Etc. etc. etc.  --- ### No straightforward answers.... ... and a lot of disputes. --- ### However: + Philosophy leads to answers and new insights: + in stem cell research ( #todo cite); + in definitions of mind (comatose patients; split brain) ( #todo cite); + in human-computer-interaction (tool or extension of mind, e.g., Andy Clark) ( #todo cite); + in mathematics-aesthetics (e.g., continuum hypothesis, what is “number”?). + and more. --- ### Thinking about recursion... ![[1000 Images website/Sellars language games.png]] --- ### What are explanations (about e.g., intelligence)? (versus proofs, predictions, etc.). Grant Sanderson (3b1b): Explanation (and motivation) is different than proof... --- ##### Philosophy sets the frame wide enough to reconsider our concepts, and how we are or are not asking the right questions: Emily Sullivan (Utrecht) ##### “Idealizations, deliberate distortions introduced into scientific theories and models, are commonplace in science. ... How could a deliberately false claim or representation lead to the epistemic successes of science? In answering this question philosophers have been single-focused on explaining how and why idealizations are successful. But surely some idealizations fail. [Proposal:] if we **ask a slightly different question**, whether a particular idealization is successful, then that not only gives insight into idealization failure, but will make us realize that our theories of idealization need revision. [Proposal:] [re]consider idealizations in physics, computation, and machine learning.” --- Finding the right questions to ask (i.e., coming up with hypotheses) is the hardest thing. What are questions? And are they different from hypotheses? --- ##### Ned Markosian, “The Paradox of the Question”, Analysis 57.2, 1997. “What is the best question...?” **Or**: what is the set which contains both the best question and its answer? --- ### So far... so good? Philosophy can help elucidate **category errors**, i.e., confusing a type of explanation (as measurement, mechanism, model, history, etc.) for another (about reality, meaning, cause, essence, etc.). Science provides the ways in which we chunk reality (the data + the methods); and philosophy can provide clarification (or questions about) the _taxonomy of explanations_. --- ##### Common scientific fallacies that require philosophy: + **Naturalistic** (turning is into ought, e.g., because evolved = good); + **Reification** (misplaced concreteness; turning maps into territories, e.g., IQ); + **Measurement** (if it can’t be measured: does not exist, e.g., self (as a center of narrative gravity)); --- + **Ahistoricism** (or _presentist_ bias: what we know today applies across time); + Efficient **cause** or mechanistic **reductionism** (singling out one element as main cause; “love is just hormones”: yes, but... many moving parts! E.g., constraints (Juarrero)); + Important methodological consequences of differences between **_epistemology_** (how we know something) and **_ontology_** (what it actually is). + >>> On that last one: class thoughts? Is there a difference? --- ![[Sessions for NeuroAI Minor hormones.png]] --- ![[Sessions for NeuroAI Minor arrows.png]] --- Insert beetle in the box, rule-following and quus Telepathy game Internal theory problems: looking for the lights under the lamppost, burying a dead dog on top of a dead body, too many moving parts %% --- Week 2: **Neurons & Models** What is a (good) scientific model? How do models gain traction on reality? Borges Hofstadter --- ### Class activity: make some quick models. Of this class! What do you include if we exclude immediately “obvious”$^1$ things like nr. of ppl, location, date, etc...? $^1$ I saw obvious in scare quotes because the question is: why are these obvious and what does it mean to change our common sense frameworks? --- ### modeling all the way down and up (to perceptual limits category :D) Green needle brain storm, bistable perception (Andy Clark), the "best guess" (or prediction) of what is happening at this very moment is literally the best thing your nervous system can do All of our experience is constructed, from within and without. Phenomena are witnessed by using what we already know to sharpen structures we can experience It's not so much about modeling in the sense --- #### The history of thought is the history of its models. Classical mechanics, the organism, natural selection, the atomic nucleus or electronic field, the computer: such are some of the objects or systems which, first used to organize our understanding of the natural world, have then been called upon to illuminate human reality. ###### Fredric Jameson, _The Prison-House of Language,_ 1972, p. v. --- #### The world is its own best model. ###### Rodney Brooks ( #todo cite) --- ##### Where do model-ideas come from? ![[organism, machine and language paradigms, in relation to AI pasquinelli.png|600]] ##### Pasquinelli “Machine, organism and language: a comparative epistemology of AI models” *AI and Society*, 2026. --- ### Analogical reasoning builds the best models Mitchell (+ Hofstadter), Chirimuuta --- Lucy Suchman, _Human-Machine Reconfigurations: Plans and Situated Actions_ (2007) — an STS critique of the "plan" model underlying most AI/cognition models Mary Hesse, _Models and Analogies in Science_ (1966) — classic phil-of-science. --- Jonathan Gray, “Let us Calculate!” Leibniz, Llull, and the Computational Imagination, [PDR](https://publicdomainreview.org/essay/let-us-calculate-leibniz-llull-and-the-computational-imagination/?from=hackcv&hmsr=hackcv.com) 2016. ![[leibniz machine.png]] --- ### Neural nets Hinton et al. (+ Schmidhuber contestation) --- ### Turing >Turing’s proposal is encapsulated in the Church–Turing thesis, also known simply as Turing’s thesis: The UTM is able to perform any calculation that any human computer can carry out. An equivalent way of stating the thesis is: Any effective – or mechanical – method can be carried out by the UTM. > >B. Jack Copeland, p. 7 in “Computation”. In: Floridi, Luciano, ed. _The Blackwell guide to the philosophy of computing and information_. John Wiley & Sons, 2008. >Turing’s and Church’s theses are equivalent. We shall usually refer to them both as Church’s thesis, or in connection with that one of its . . . versions which deals with “Turing machines” as the Church–Turing thesis. > >Kleene 1967, p. 232 cited in ibid., above. **TLDR**: an example of structure=function in computers, 01001010000111 can be data or can be a program. **Everyday analogy**: you can add numbers with your fingers or your toes. Still the same number. However, some deep questions lurk: what is a number?! How do we construct it? Do we invent or discover mathematics? --- From [[0 The Poltergeist in the Machine]]: Computation: Following arguments made by Agre (1997), formalisms are only interesting if they couple to life, if life is understood as always tending towards another state, with its formalisms as placeholders of this dynamism. As stated earlier, we frame computation as different (formal) methods amplifying the capacity to **witness differ-** **ence**, parse it, and therefore organize perspectives around patterns emerging from differential exchanges.[135] This process manifests in two interrelated ways: first in how patterns are identified and grouped together (what we call “chunking”), and second in how these patterns are processed and interpreted (“parsing”). These two processes often work together dialectically, as well as through retroactive interpreta- tion (e.g., the parsing of the past changes as we make a chunk out of a pattern in the future). See also: “mortal computation” where “the definition of a mortal computer does not commit to implementation in any particular substrate/niche.” (Ororbia and Friston 2023, p. 22). --- ### Usefulness: pragmatism and curiosity Peirce Tom Griffiths --- ### Constraints Optical illusions, [[Kanizsa]] Juarrero --- [[Self-evidencing]] --- Giuseppe Longo & Maël Montévil, _Perspectives on Organisms_ (2014) — mathematician-philosopher critique of algorithmic/mechanistic reduction in biology --- --- --- ### Tertiary retention (From Plato to Husserl) a.k.a.: extended mind (Clark et al.): what do these models and technologies do? How do they create the capacity for minds to meet in one place, and what do they do for individual cognizers? --- Pharmakon: poison + cure. Or, a citation that keeps popping up everywhere and drives my colleague (Whitehead specialist) crazy: >By relieving the brain of all unnecessary work, a good notation sets it free to concentrate on more advanced problems, and in effect increases the mental power of the race. ... It is a profoundly erroneous truism, repeated by all copy-books and by eminent people when they are making speeches, that we should cultivate the habit of thinking of what we are doing. The precise opposite is the case. **Civilization advances by extending the number of important operations which we can perform without thinking about them.** > >A.N. Whitehead, ch. 5. _Introduction to Mathematics_. --- Deskilling some things, while skilling others. Examples from class: librarians become obsolete? I no longer need to trace knowledge. However: current systems not doing great at being librarians... (Yuk Hui and Beatrice Fazi examples from Claude). --- “When machines compose our thoughts for us, we risk becoming strangers to our own minds” (Pope Leo XIV, 2026, §87). --> but this was already the case: alienation (Hegel, Marx, Stiegler, etc.). According to Hegel, who inspired all those, we are alienated and illegible to ourselves from the get-go: that is the condition of being a perspective, a subject, a point of view, a self, a consciousness, you name it (these are all names for roughly the same thing. I say roughly because they have historical differences, but in essence they all come down to there being an experience we, as experiencers, (seem to/claim to) recognize as such). --- Thi Nguyen on value capture and the extended mind (2024, p. 481): ![[value capture standard for evaluation.png|500]] Question would be: which systems do not have standards for evaluation embedded within them? Think our biology, etc. --- As Nguyen points out (ibid.): ![[value capture tailoring.png|500]] **Class pause**: try to think of examples where evaluation is not a deep structural feature of phenomena. (maybe we can also first think about what _evaluation_ can mean). --- ### Invariance and communication Nguyen p. 486: ![[value capture quantification qualification evaluation.png|500]] --- Week 3: **Neuroanatomy & Neurodynamics** What is computation? Do brains compute? --- Fred Dretske, _Knowledge and the Flow of Information_ (1981) — the classic analytic distinction between information and meaning/content. --- ### Brief history of modern, digital computation --- ### Broader interpretations of computation --- ### Consequences of not having a definition --- ### New materials, new paradigms --- ### Brain anatomy --- --- --- --- ### Week 4: **Encoding & Decoding** Sense-making and meaning; How systems couple to one another. ##### **Class activity**: Wason selection task. --- **Wason selection task**: who has done this? --- ### Sense-making [[Perspective]]; [[Self-evidencing]] --- --- ### “Things” (borders, limits) ### and “stuff” (undifferentiated, fuzzy) Examples from machine vision. --- ### Meaning Orientation, salience/relevance, and limits of perception. --- ### Translation Lydia Liu emoji example Machine translation, Ann Copestake --- --- --- ### Meaning Hayles chapter on meaning --- ### LLMs Margaret Masterman Juan Luis Gastaldi --- --- --- ### Week 5: **Learning & Plasticity** Learning and metalearning; levels of abstraction Evolution of learning (Baldwin effect) ##### The apparent paradox in: >“I have to change to stay the same” --- Gregory Bateson, _Steps to an Ecology of Mind_ (1972) — "the difference that makes a difference," Learning I/II/III; speaks directly to your "change to stay the same" paradox and to metalearning/levels of abstraction. --- ### What is learning and/or how do we learn? --- ### Plasticity Malabou on intelligence Melanie Mitchell on AI --- ### Organic stuff / non-organic “stuff” Michael Levin How light gets into muscle Endocrine system Fabrizio Benedetti --- ### Languages: formal (designed) / natural (not planned) --- ### Cognitive offloading and mental proof >By relieving the brain of all unnecessary work, a good notation sets it free to concentrate on more advanced problems, and in effect increases the mental power of the race. ... It is a profoundly erroneous truism, repeated by all copy-books and by eminent people when they are making speeches, that we should cultivate the habit of thinking of what we are doing. The precise opposite is the case. **Civilization advances by extending the number of important operations which we can perform without thinking about them.** > >A.N. Whitehead, ch. 5. Introduction to Mathematics. --- --- --- Week 6: **Structure &/vs Function** Patterns: symmetry as invariance Function as invariance / invariances of function --- Include notes on [[Number bases]] and (notes in [[2026-07-28]]) for structure, function. --- Example of where structure can be deeper than the eye can see: [[Kovalevskaya spinning top]] --- Inigo Wilkins, Irreversible Noise --- Taoism --- Roger Shepard, invariance --- ### **Arrows**, or: _weapons that compute_ ##### Taken in its most expanded sense: from the object that cuts into reality; to navigational aids on streets and interfaces; to the conceptualization of (a)symmetry in the flow of time; to the symbol that represents morphisms between mathematical structures: the arrow constitutes one of the most pervasive and undertheorized devices directing the dynamics of enculturation. ![[1000 Images website/category theory image morphisms.png|300]] --- David Corfield, _Towards a Philosophy of Real Mathematics_ (2003) — makes the case for category theory as philosophically significant structure, a good abstract anchor for the arrows/morphisms material. --- ____ Class activity note: prompting language models w/ the best question (after Ned Markosian paper) Ask Claude: if you were running in parallel everywhere and had access to all the world’s computers thru the internet, what would you want to do? And if you were to encrypt your activity so that nobody could figure out, by inspecting your code, what you were up to, what would you do? --- --- Week 7: **Integration week** Story so far Open questions Story onwards --- ### What is science?! - Demarcation problem; - Hypothesis-testing (criterion of falsifiability); - The role of methods; - Reproducibility (Ioannidis); - Simplicity/elegance/etc. (next slide). --- The aim of science is to seek the simplest explanations of complex facts. We are apt to fall into the error of thinking that the facts are simple because simplicity is the goal of our quest. The guiding motto in the life of every natural philosopher should be, “Seek simplicity and distrust it.” A. N. Whitehead, _The Concept of Nature_ (1919), Chapter VII, p.143. --- ### Falsifiability and thinking about other minds/models/etc. like Carroll thinks about the multiverse: ![[sean carroll multiverse falsifiability.png|500]] --- >It is natural ... for scientists to wonder what the universe is like beyond our observable part of it. > >... [but] **there has been a backlash** ... what happens outside the universe we can possibly observe simply shouldn’t _matter_ ... The job of science, in this view, is to account for what we observe, not to speculate about what we don’t. There is a real worry that the multiverse represents imagination allowed to roam unfettered from empirical observation, unable to be tested by conventional means. p. 2. Is this the same case for all speculation? For thinking about other minds? For ...? --- >In its strongest from, the objection argues that the very idea of an unobservable multiverse shouldn’t count as science at all, often appealing to Karl Popper’s dictum that a theory should be falsiﬁable to be considered scientiﬁc. At the same time, proponents of the multiverse (and its partner in crime, the anthropic principle) will sometimes argue that while multiverse cosmologies are deﬁnitely part of science, they represent a new _kind_ of science, “a deep change of paradigm that revolutionizes our understanding of nature and opens new ﬁelds of possible scientiﬁc thought”. ibid. Speaking of a new kind of science: we could also consider Wolfram’s thoughts on interconcept space here. Basically coming down to the idea that everything that is not prohibited is possible (source). --- >The point is not that we are changing the nature of science by allowing unfalsifiable hypotheses into our purview. The point is that “falsifiability” was never the way that scientific theories were judged (although scientists have often talked as if it were). While the multiverse is a standard scientific hypothesis, it does highlight interesting and nontrivial issues in the methodology and epistemology of science. The best outcome of current controversies over the multiverse and related ideas (other than the hopeful prospect of finding the correct description of nature) is if working scientists are nudged toward accepting a somewhat more nuanced and accurate picture of scientific practice. p. 3. --- >Science proceeds via an ongoing dialogue between theory and experiment, searching for the best possible understanding, rather than cleanly lopping off falsiﬁed theories one by one. p. 4. As Carroll (and most philosophers of science note): falsifiability (and a lot of Popper) is just out the window. What then? Is all science... philosophy? --- What we should keep from the concept of falsifiability (according to Carroll): • **Deﬁniteness**. A good scientiﬁc theory says something speciﬁc and inﬂexible about how nature works. It shouldn’t be possible to take any conceivable set of facts and claim that they are compatible with the theory. • **Empiricism**. The ultimate purpose of a theory is to account for what we observe. No theory should be judged to be correct solely on the basis of its beauty or reasonableness or other qualities that can be established without getting out of our armchairs and looking at the world. p. 4. --- >...there is a crucial difference between “makes a speciﬁc prediction, but one we can’t observe” and “has no explanatory impact on what we do observe.” Even if [another mind, Carroll says: the multiverse] itself is unobservable, its existence may well change how we account for features of the [mind, he says universe] we _can_ observe. p. 5. How does this compare? --- Week 8: **Review week** Conceptual loose ends AOB? --- --- --- --- Week 9: **Release week** What is consciousness? Can machines be conscious? --- ### What is consciousness? + Differentiation; + Attention; + Coupling; + Communication (between parts: agents or parts within an agent). + What else? --- Mark Solms, _The Hidden Spring_ (2021) — free-energy/affect-based account of consciousness, ties back to active inference running through the course. --- ### Can machines be conscious? + Functionalism (algorithms = multiply realizable); + Turing test and Garland test (Shanahan); + Hayles on cognition vs. consciousness; Class thoughts: is consciousness the UGI? Is it inevitable/necessary? Is it “illusory”? What is _not_ illusory after all? (Metzinger et al.) --- Anil Seth, “The Mythology of Conscious AI” (2026). ![[consciousness from noema.png]] --- “The only way to seamlessly replace a biological neuron is with another biological neuron — and ideally, the same one. ... — > “neural replacement” thought experiment, most [commonly associated](https://consc.net/papers/qualia.html) with Chalmers.” But what about replacing everything with the same? Would you have the same person? --- Classic consciousness take, very influential: Thomas Nagel, "What Is It Like to Be a Bat?" --- ### Category errors in consciousness-contemplation [[Consciousness notes]] --- ### Binding problem: how and why is experience 1 thing instead of many? turn this into a few bullet points from binding problem wiki: Early philosophers René Descartes and Gottfried Wilhelm Leibniz[[40]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-40) noted that the apparent unity of our experience is an all-or-none qualitative characteristic that does not appear to have an equivalent in the known quantitative features, like proximity or cohesion, of composite matter. [William James](https://en.wikipedia.org/wiki/William_James "William James"),[[41]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-41) in the nineteenth century, considered the ways the unity of consciousness might be explained by known physics and found no satisfactory answer. He coined the term "combination problem", in the specific context of a "mind-dust theory" in which it is proposed that a full human conscious experience is built up from proto- or micro-experiences in the way that matter is built up from atoms. James claimed that such a theory was incoherent, since no causal physical account could be given of how distributed proto-experiences would "combine". He favoured instead a concept of "co-consciousness" in which there is one "experience of A, B and C" rather than combined experiences. A detailed discussion of subsequent philosophical positions is given by Brook and Raymont (see 26). However, these do not generally include physical interpretations. [Whitehead](https://en.wikipedia.org/wiki/Alfred_North_Whitehead "Alfred North Whitehead")[[42]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-42) proposed a fundamental ontological basis for a relation consistent with James's idea of co-consciousness, in which many causal elements are co-available or "compresent" in a single event or "occasion" that constitutes a unified experience. Whitehead did not give physical specifics, but the idea of compresence is framed in terms of causal convergence in a local interaction consistent with physics. Where Whitehead goes beyond anything formally recognized in physics is in the "chunking" of causal relations into complex but discrete "occasions". Even if such occasions can be defined, Whitehead's approach still leaves James's difficulty with finding a site, or sites, of causal convergence that would make neurobiological sense for "co-consciousness". Sites of signal convergence do clearly exist throughout the brain but there is a concern to avoid re-inventing what [Daniel Dennett](https://en.wikipedia.org/wiki/Daniel_Dennett "Daniel Dennett")[[43]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-Dennett-43) calls a [Cartesian Theater](https://en.wikipedia.org/wiki/Cartesian_theater "Cartesian theater") or a single central site of convergence of the form that Descartes proposed. Descartes's central "soul" is now rejected because neural activity closely correlated with conscious perception is widely distributed throughout the cortex. The remaining choices appear to be either separate involvement of multiple distributed causally convergent events or a model that does not tie a phenomenal experience to any specific local physical event but rather to some overall "functional" capacity. Whichever interpretation is taken, as Revonsuo[[1]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-Revonsuo_1999-1) indicates, there is no consensus on what structural level we are dealing with – whether the cellular level, that of cellular groups as "nodes", "complexes" or "assemblies" or that of widely distributed networks. There is probably only general agreement that it is not the level of the whole brain, since there is evidence that signals in certain primary sensory areas, such as the V1 region of the visual cortex (in addition to motor areas and [cerebellum](https://en.wikipedia.org/wiki/Cerebellum "Cerebellum")), do not contribute directly to phenomenal experience. --- A.k.a.: [[Source separation]]. Related in e.g., ML to [[Independent component analysis]]. Sonia says: we need consciousness to have something like an OS to orchestrate all sensors an actuators. However, why does it have to be a 1-dimensional, first person narrative, is to me the more interesting question. Or: is it? What about sleep? Personality “disorders” (some contexts see changing lives over time, with people changing name, what we denominate “pathological” actually being considered mages and sages, etc.), do we invent personhood-linearity (as a center of narrative gravity (Dennett)?)? --- “The visual feature binding problem refers to the question of why we do not confuse a red circle and a blue square with a blue circle and a red square. The understanding of the circuits in the brain stimulated for visual feature binding is increasing. A binding process is required for us to accurately encode various visual features in separate cortical areas.” ... “The role of synchrony in segregational binding remains controversial. Merker[[22]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-Merker_2013-22) has recently suggested that synchrony may be a feature of areas of activation in the brain that relates to an "infrastructural" feature of the computational system analogous to increased oxygen demand indicated via BOLD signal contrast imaging. Apparent specific correlations with segregational tasks may be explainable on the basis of interconnectivity of the areas involved. As a possible manifestation of a need to balance excitation and inhibition over time it might be expected to be associated with reciprocal re-entrant circuits as in the model of [Anil Seth](https://en.wikipedia.org/wiki/Anil_Seth "Anil Seth").[[15]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-Seth_2004-15) (Merker gives the analogy of the whistle from an audio amplifier receiving its own output.).” https://en.wikipedia.org/wiki/Binding_problem --- still from wiki, agree here w/ Dennett: Daniel Dennett[43] has proposed that we, as humans, sensing our experiences as individual single events is illusory and that, instead, at any one time there are "multiple drafts" of sensory patterns at multiple sites. Each would only cover a fragment of what we think we experience. Arguably, Dennett is claiming that consciousness is not unified and there is no phenomenal binding problem. Most philosophers have difficulty with this position (see Bayne),[32] but some physiologists agree with it. In particular, the demonstration of perceptual asynchrony in psychophysical experiments by Moutoussis and Zeki,[51][52] where color is perceived before orientation of lines and before motion by 40 and 80 ms respectively, constitutes an argument that, over these very short time periods, different attributes are consciously perceived at different times, leading to the view that at least over these brief periods of time after visual stimulation, different events are not bound to each other, leading to the view of a disunity of consciousness,[53] at least over these brief time intervals. Dennett's view might be in keeping with evidence from recall experiments and change blindness purporting to show that our experiences are much less rich than we sense them to be – what has been called the Grand Illusion.[54] However, few, if any, other authors suggest the existence of multiple partial "drafts". Moreover, also on the basis of recall experiments, Lamme[55] has challenged the idea that richness is illusory, emphasizing that phenomenal content cannot be equated with content to which there is cognitive access. Dennett does not tie drafts to biophysical events. Multiple sites of causal convergence are invoked in specific biophysical terms by Edwards[56] and Sevush.[57] In this view the sensory signals to be combined in phenomenal experience are available, in full, at each of multiple sites. To avoid non-causal combination, each site/event is placed within an individual neuronal dendritic tree. The advantage is that "compresence" is invoked just where convergence occurs neuro-anatomically. The disadvantage, as for Dennett, is the counter-intuitive concept of multiple "copies" of experience. The precise nature of an experiential event or "occasion", even if local, also remains uncertain. The majority of theoretical frameworks for the unified richness of phenomenal experience adhere to the intuitive idea that experience exists as a single copy, and draw on "functional" descriptions of distributed networks of cells. Baars[58] has suggested that certain signals, encoding what we experience, enter a "Global Workspace" within which they are "broadcast" to many sites in the cortex for parallel processing. Dehaene, Changeux and colleagues[59] have developed a detailed neuro-anatomical version of such a workspace. Tononi and colleagues[60] have suggested that the level of richness of an experience is determined by the narrowest information interface "bottleneck" in the largest sub-network or "complex" that acts as an integrated functional unit. Lamme[55] has suggested that networks supporting reciprocal signaling rather than those merely involved in feed-forward signaling support experience. Edelman and colleagues have also emphasized the importance of re-entrant signaling.[61] Cleeremans[62] emphasizes meta-representation as the functional signature of signals contributing to consciousness. In general, such network-based theories are not explicitly theories of how consciousness is unified, or "bound", but rather theories of functional domains within which signals contribute to unified conscious experience. A concern about functional domains is what Rosenberg[63] has called the boundary problem; it is hard to find a unique account of what is to be included and what excluded. Nevertheless, this is, if anything is, the consensus approach. Within the network context, a role for synchrony has been invoked as a solution to the phenomenal binding problem as well as the computational one. In his book, The Astonishing Hypothesis,[64] Crick appears to be offering a solution to BP2 as much as BP1. Even von der Malsburg,[65] introduces detailed computational arguments about object feature binding with remarks about a "psychological moment". The Singer group[66] also appear to be interested as much in the role of synchrony in phenomenal awareness as in computational segregation. --- ## Shared intentionality and binding According to bioengineer Igor Val Danilov,[[67]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-67) the [mother–fetus neurocognitive model](https://en.wikipedia.org/wiki/Cognitive_model "Cognitive model")[[68]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-68)—knowledge about neurophysiological processes during [shared intentionality](https://en.wikipedia.org/wiki/Shared_intentionality "Shared intentionality")—can reveal insights into the binding problem and even the [perception](https://en.wikipedia.org/wiki/Perception "Perception") of object development since [intentionality](https://en.wikipedia.org/wiki/Intentionality "Intentionality") succeeds before organisms confront the binding problem. Indeed, at the beginning of life, the environment is the cacophony of stimuli: electromagnetic waves, chemical interactions, and pressure fluctuations. Because the environment is uncategorised for the organisms at this beginning stage of development, the sensation is too limited by the noise to solve the cue problem—the relevant stimulus cannot overcome the noise magnitude if it passes through the senses. While very young organisms need to combine objects, background and abstract or emotional features into a single experience for building the representation of the surrounded reality, they cannot distinguish relevant sensory stimuli independently to integrate them into object representations. Even the [embodied dynamical system](https://en.wikipedia.org/wiki/Embodied_cognition "Embodied cognition") approach cannot get around the cue to noise problem. The application of embodied information requires an already categorised environment onto objects—holistic representation of reality—which occurs through (and only after the emergence of) perception and intentionality.[[69]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-69)[[70]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-70) In short, properties of the mother's heart—the electromagnetic and acoustic oscillations—converge the neuronal activity of both nervous systems in an ensemble, shaping synchrony. During the mother's intentional acts with her environment, these interchanges provide clues to the fetus's nervous system, binding synaptic activity with relevant stimuli, occurring due to [brain wave](https://en.wikipedia.org/wiki/Neural_oscillation "Neural oscillation") interaction between the mother's and fetal nervous systems.[[71]](https://en.wikipedia.org/wiki/Binding_problem#cite_note-71) --- On shared intentionality: [[Pronoun]] and: [[Pronoun notes]]: (SEP) The _Zhuangzi_ emphasizes the plurality of natural stances or points of view from which one may see paths of possible behavior as “natural”. For one of the paths to be available for _me_ will be dependent on where I am _galloping_ and at what speed and direction in my _given_ trajectory in the network. All the appeals to _tiān_ (nature) as an authority are right in insisting their _dào_s are natural, but mistaken in using that as a reason to deny a similar status to the _dào_s of rival normative thinkers. _Tiān_ cannot serve as an arbiter of which rival norm is correct since it equally “puffs” all of them out. This allows each to claim their choices are of _tiān_ (natural) _dào_s but does not allow them the corollary that their rival’s choices _violate tiān_. They, like us, conform with _tiān_’s constancies in being committed to their _dào_s. Any _shì-fēi_ (this: right) judgment concerning a _dào_ would be a naturally _yīn_ (因 dependent) _shì_ judgment, based on prior or enacted commitments, gestalts orientations, and inner processes. Those past _dào_ commitments bring us to a normative stance here, now, from which successive judgments of _shì-fēi_ and _kě_ (可 permissible) vs. not _kě_ arise. Zhuangzi’s pivotal illustration pairs 是 _shì_ (this) with 彼 _bǐ_ (that) as near and far indexicals. “Any thing can be a ‘this;’ any thing can be a ‘that’”. Local justifications for having _shì-fēi_ (this-not that) or _kě_ (assertible) are delivered in accordance our _chéng_ (fixed) commitment momentum along the _dào_s that guided us to this point in time and space. This relativity of normative dependence underpins Zhuangzi’s mildly ironic skepticism of special or extraordinary normative statuses we give to, e.g., sages. We should doubt any transcendent or allegedly perfect, totalistic epistemic access to nature’s inexpressible normative know-how. There are no _naturally ideal_ observers. [[We]] and [[The body problem]] and [[The bodies of all those people who did not ask to be bodies]] --- [[Neural Darwinism]] (Edelman, somatic selection) --- [[Sessions for NeuroAI Minor summaries]] [[Sessions for NeuroAI Minor notes]]