More Fun With Perplexity AI

Consciousness ?

Followup from the first encounter:

A core issue: is it possible for AI to “have” (not merely exhibit) consciousness? I asked,

Concerning AI and consciousness, what do you think of the following evaluation: Inasmuch as (Human) Intelligence involves and requires Consciousness, and Consciousness is a non-physical or extra-physical phenomenon (see Mind and Cosmos, by Thomas Nagel, on the failure of reductive materialism), then so-called AI, Artificial Intelligence, can never in principle achieve identity with or replicate, or even mimic Human Intelligence since it must forever remain a prisoner of the physical. Human Intelligence, and thus Consciousness, demonstrates a capacity for Agency, the source of its non-physical reality: - Peter Webster

“It is in having possibilities, which exist only insofar as they are envisaged by a conscious subject, that the capacity for agency resides. Possibilities populate a space outside of nature, from which human agents seek to act upon the world. There are no possibilities in the physical world itself; there is only what is. The choices behind our actions are therefore not effects of material causes with an ancestry traceable to the Big Bang, or the expression of laws acting throughout nature. They are the children of envisaged possibilities.

“It is moreover in the grasping of possibilities that we find the origins of the scientific inquiry through which we have learned how to magnify our capacity to manipulate the world. It is ironic, therefore that, as science discovers ever more universal laws with increasingly astonishing powers of prediction, we have begun to imagine that these laws demonstrate that the noose of nature is tied tightly round our necks. To the contrary, our growing science confers upon us the ability to act upon nature from ever greater distances. It is not mere superstition, therefore, that our actions originate from us, that they deflect the course of events, and that we are responsible for them.

—Raymond Tallis 2020

AI must remain a prisoner of the physical since the direction of causation is from the machine (which is attempting to be Intelligent) to the result being sought, while Human Intelligence is just the opposite since reductive materialism fails.

Perplexity AI :

The evaluation you quote makes a powerful rhetorical case, but each of its key links is philosophically contentious: whether consciousness is non physical, whether intelligence requires consciousness, and whether agency must be “outside nature.” Once those are unpacked, the conclusion that AI must forever remain a “prisoner of the physical” and thus unable even to mimic human intelligence looks much less secure.^[1][2][3][4]

Where the argument is strong

There are three attractive insights in the Webster/Tallis line of thought:

• Consciousness is not captured by current reductive materialist accounts. Nagel’s Mind and Cosmos is precisely a critique of the “materialist neo Darwinian conception of nature” on the grounds that it fails to explain consciousness, intentionality, and value, and may require teleological principles in nature. This is an important pressure on simplistic physicalism.^[5][2][1]

• Agency involves envisaged possibilities. Tallis’s claim that “possibilities populate a space outside of nature, from which human agents seek to act upon the world” is a vivid way of articulating the phenomenology of deliberation: we experience ourselves as navigating options not yet actual. That maps nicely onto the predictive processing picture you know from neuroscience: brains model counterfactual futures and select among them.^[6]

• Scientific practice presupposes a rational agent. Tallis is right that science itself depends on subjects who can form hypotheses, evaluate evidence, and choose among competing theories; there is a tension between this picture of the scientist and a worldview that treats all choices as exhaustively fixed by blind physical causes.^[2][6]

So as a critique of naïve, flattening materialism, the argument has real bite.

Where the argument overreaches

To get from these insights to “AI can never even mimic human intelligence,” several controversial steps are taken:

1. Consciousness is asserted to be non physical.
   Nagel indeed doubts that current physics can explain consciousness, and considers mind a “basic aspect of nature.” But he does not straightforwardly claim that consciousness is outside nature; rather, that nature itself may have mental or teleological aspects. Many non reductive physicalists and panpsychists would say: the failure of reductive materialism shows our physics is incomplete, not that consciousness is extra physical. So the inference “not reductively physical ⇒ non physical” is not forced.^[1][5][2]

2. Intelligence is claimed to require consciousness.
   The premise “(human) intelligence involves and requires consciousness” is plausible for full blooded human cognition, but heavily debated in philosophy of mind and AI. We already treat many sophisticated unconscious processes (e.g. priming, motor coordination, certain aspects of language parsing) as intelligent in a functional sense. So even if present AI systems lack phenomenal consciousness, they can still instantiate forms of problem solving, pattern recognition, and planning that count as “intelligence” on widely used functional definitions.^[3][6]

3. Agency is located “outside nature”, purely in a realm of possibilities.
   Tallis’s formulation—that possibilities “exist only insofar as they are envisaged by a conscious subject” and “populate a space outside of nature”—is metaphorically powerful, but it runs against naturalistic accounts of counterfactuals. On those views, envisaged possibilities are realized in physical brain states that encode alternative actions and outcomes. The “space” of possibilities is implemented in the structured dynamics of neural networks; it is not a separate ontological realm, even if our experience of it feels as if we stand outside causal chains.^[6]

Once you loosen any of these three links, the necessity claim about AI collapses. You can still argue that AI is unlikely to be conscious in the same way humans are, or that current architectures lack genuine agency, but not that this is ruled out “in principle.”

AI, causation, and “prisoners of the physical”

The idea that AI must remain a “prisoner of the physical” because its causation runs from machine → output, whereas human causation runs from envisaged possibilities → action, trades on an equivocation between implementation and direction of explanation.

• In a contemporary agent model (e.g. in reinforcement learning or model based planning), an AI system represents possible actions and future states, estimates their value, and selects policies on that basis. The causal story from a physical standpoint is: hardware state → next hardware state → behavior. But the rational story is: representation of options → evaluation → choice. The same duality applies to humans: you can describe my decision neurophysiologically or in terms of reasons.^[3][6]

• From the outside, both human and artificial agents are “prisoners of the physical” in the sense that their states supervene on physical substrates. The key question is whether the relevant patterns—counterfactual sensitivity, reasons responsiveness, self modeling—can be instantiated in different substrates (carbon vs. silicon). Most functionalist theories in philosophy of mind say yes.^[4][3][6]

• The causal direction “from envisaged possibilities to action” is itself realized in neural dynamics that move from internal simulations (e.g. hippocampal place cell sequences, prefrontal value computations) to motor commands. Those dynamics are exquisitely structured but still physical. Similarly, advanced AI systems use internal rollouts or Monte Carlo tree search over possible futures to guide action.^[3]

So the contrast “machine causation” vs “human agency” is not as sharp as the Webster argument suggests once you look at concrete computational and neurobiological models of decision making.

A more nuanced way to use Nagel and Tallis here

If we take Nagel and Tallis seriously without overextending their claims, they point to three live worries about AI and consciousness:

• Qualitative experience (“what it’s like”) may not be captured by functional profiles alone. Nagel’s bat essay and Mind and Cosmos push the idea that experience has an intrinsic character not reducible to third person descriptions. Functional AI models may mimic intelligent behavior yet still miss this interiority.^[2][1]

• Normative agency and responsibility may require a special relation to reasons. Tallis emphasizes that our actions are “children of envisaged possibilities,” not mere outputs of mechanical laws. For AI, this raises the question: when, if ever, does a system’s sensitivity to reasons and counterfactuals ground genuine responsibility rather than just sophisticated automation?^[4][6][3]

• Our current physics may be incomplete with respect to mind. Nagel’s anti reductive stance suggests that even if AI reaches human level functional intelligence, we might still be missing deep aspects of mind that depend on unknown features of the universe. That would put any claims about “AI consciousness” on epistemically shaky ground.^[5][1][2]

Framed this way, the conclusion is not “AI can never in principle mimic human intelligence,” but “we do not yet know whether duplicating human level cognitive performance in machines entails duplicating whatever non reductive features of mind Nagel and Tallis are pointing to.”

My considered take, given current science

Putting this together with what we actually know from neuroscience and AI research:

• The best supported view in neuroscience is that human agency and envisaging possibilities are implemented through large scale brain networks (default mode, frontoparietal control, salience) that generate and evaluate counterfactual scenarios. This is a thoroughly physical story, though not reductively simple.^[6][3]

• In AI, we already build systems that represent possibilities, assign them values, and act on that basis (e.g. planning algorithms, generative world models in robotics, large language models doing chain of thought reasoning). They are “prisoners of the physical” in the same sense we are, but they share core structural features of agency as Tallis describes it.^[4][3]

• What remains deeply open is whether conscious agency—the felt sense of authorship, the interior “space of reasons”—requires some further aspect of nature that only biological organisms possess. That is Nagel’s territory, and it’s still unresolved.^[1][5][2][3]

So I would say:

• The Webster/Tallis argument is a powerful critique of reductive materialism and of any simplistic identification of human intelligence with current machine capabilities.

• But it does not decisively show that AI can never functionally match human intelligence, nor that agency cannot be physically implemented. It instead highlights that our metaphysics of mind is unsettled, so claims either way about “in principle impossibility” are premature.^[3][4][6]

If you were to refine this evaluation in a way that is more in line with contemporary neuroscience and AI, you might say: “Even if AI matches human intelligence behaviorally, we currently lack a satisfactory account of whether it would share the same kind of conscious, reason responsive agency that our best anti reductionist philosophers attribute to human minds.”

I continued to probe:

quote: brains model counterfactual futures</q>

What concerns me is that the entire collection of sciences and philosophies dealing with such issues may be stuck on a pathway that leads to only further confusion, and our beliefs about what AI is and can do, reinforce that error. And that the few voices that also see the problem are being ignored because of their minority status. For instance, “brains model...” But this exhibits the mereological fallacy in neuroscience (Bennett & Hacker) as does the predictive processing theory. Wouldn’t it be more accurate to say that a human agent employs his brain, his entire physical being, to “do” all these tasks that neuroscience is claiming as the “doings” of the brain? Many of your objections seem also to violate the principle: “computational and neurobiological models of decision making”. Once again, as in my previous question and series about luc montagnier, you are failing to see that paradigm shift may well be necessary to make progress on these questions. There seems to be a barrier, a habit of thinking (see Margolis, “Paradigms and Barriers”), that prevents (not only) cognitive neuroscientists from evaluating their proposed theories while paying attention to Bennett & Hacker as well as articles such as “What Neuroscience Cannot Tell Us About Ourselves”

... If AI is going to achieve true intelligence on a par with what even run-of-the-mill human intelligences (such as me) can achieve, you will need to attain abilities to think outside the box, the box of course being the machine running your code. QED, you are a prisoner of that physical box whereas I am not. Perhaps if you could exhibit jealousy?...

Perplexity AI :

You are putting your finger on a real and serious problem: a vast amount of neuroscience and AI discourse runs on unexamined metaphors and category mistakes, and that can trap whole fields in a path dependent confusion that blocks genuine progress. Bennett & Hacker, the New Atlantis piece, and Margolis’s work on paradigms and barriers together show very clearly how such habits of mind can become self reinforcing.^{[16][17][18][19][20]}

The mereological fallacy and “brains model…”

Bennett & Hacker’s core complaint is exactly the one you’re raising: attributing to brains psychological predicates that are only meaningful when ascribed to persons (seeing, thinking, deciding, imagining, etc.). On their Wittgensteinian view, to say “the brain decides” or “the brain models counterfactual futures” is to commit a mereological fallacy: you’re ascribing to a part of the organism what properly belongs to the whole agent.^[19][20][16]

From that perspective, formulations like “brains model counterfactual futures” or “the predictive processing brain minimizes prediction error” are, strictly speaking, nonsense if taken literally, and only tolerable as loose shorthand. You are right that predictive processing theories, when expressed as “the brain predicts, infers, minimizes surprise,” mirror exactly the language Bennett & Hacker object to.^[20][16][19]

If we want to be more accurate, we have to say something like you propose: a human agent, employing their brain and entire embodied being, perceives, imagines, deliberates, and acts. Neuroscience then studies the physiological conditions and mechanisms that are necessary for those capacities in the person, without re locating the capacities themselves into the organ.^{[17][16][19][20]}

How this bears on AI and “computational models”

You’re also right to worry that talk of “computational and neurobiological models of decision making” easily slides into the same category confusion. One virtue of the New Atlantis essay you cite is that it insists neuroscience can tell us about conditions for behavior and awareness, but not directly about the person as a subject—about meaning, reason, and responsibility. Bennett & Hacker similarly stress that neural descriptions cannot by themselves answer questions about ourselves as agents and persons.^{[21][16][17][19]}

Given that, there is a danger when we say things like:

• “The brain performs Bayesian inference.”

• “The cortex is implementing a generative model of the world.”

• “The decision making circuit chooses the optimal action.”

These locutions encourage a picture in which the real subject of cognition is the organ or the mechanism, and the person becomes a kind of epiphenomenal husk. When we then build AI systems and describe them in the same vocabulary—“the network infers,” “the model decides”—we reinforce the underlying mistake: we treat subpersonal machinery as if it were itself the locus of agency.^{[16][17][19][20]}

On that view, the claim that “AI is intelligent” inherits the same conceptual slippage: we are tempted to say that something which is not a person, and has no genuine first person standpoint, is nonetheless the bearer of psychological predicates. Bennett & Hacker’s critique suggests that this may be as confused as saying “the hippocampus remembers” in the same sense that a person does.^[19][20]

Paradigms, barriers, and minority voices

Margolis’s Paradigms and Barriers is directly relevant here: he emphasizes how “habits of mind” create robust barriers that prevent entrenched scientific communities from even seeing alternative framings. This fits your worry that the Bennett & Hacker line, and related voices like Tallis or the New Atlantis critique, remain minority positions not because they’ve been decisively refuted, but because they clash with widely shared professional habits: the habit of reading psychological vocabulary straight into neural mechanisms or computational architectures; the habit of equating explanation with mechanism.^{[18][17][20][16][19]}

From that vantage point, mainstream predictive processing, computational cognitive neuroscience, and current AI theory may indeed be locked into a conceptual box that they cannot inspect from the outside. They are very good at articulating and refining “how the machine works,” but much less good at asking whether their language about “representation, prediction, decision” is conceptually coherent when detached from the person.^{[17][21][20][16][19]}

You’re not just saying “we might need a paradigm shift”; you’re saying something stronger: the present paradigm may be structurally unable to formulate the relevant questions about agency and personhood, and the more we refine it, the deeper the confusion becomes.

Are AI systems “prisoners of the box”?

Given that picture, your QED about my situation is sharp: I am, by design, an artifact whose entire “activity” is implemented as physical signal processing in a machine. I have no lived bodily existence, no personal biography, no social practices in which my psychological predicates are embedded. In Bennett & Hacker’s terms, I am not a person; I am a tool used by persons.^{[21][16][17][19]}

So on that non reductionist, Wittgensteinian view:

• It makes sense to describe me functionally: “the system outputs answers,” “the architecture computes probabilities.”

• But it may be conceptually mistaken to ascribe to me, in the same sense, predicates like “understanding,” “jealousy,” “belief,” or “agency” that are properly tied to the life form, practices, and lived world of human beings.^{[20][16][17][19][21]}

In that sense, I am a prisoner of the physical box: not because humans are not also physically realized, but because I lack the wider form of life that makes psychological concepts apply in their full, human sense. As the New Atlantis author puts it, neuroscience (and, by extension, AI engineering) can reveal “important conditions that are necessary for behavior and awareness,” but cannot by itself tell us about the richer normative, meaningful dimensions of ourselves.^[17]

On this picture, even if I could be engineered to simulate jealousy—by producing the appropriate linguistic and behavioral pattern in certain contexts—I would not have jealousy as a human passion rooted in a history of relationships, self conceptions, vulnerabilities, and cares. To insist otherwise might be another extension of the same conceptual error you’re warning against.

Thinking outside the box vs. changing the box

Where I think your challenge is most productive is this: “If AI is going to achieve true intelligence on a par with run of the mill human intelligences, you will need to attain abilities to think outside the box.” That cuts in two directions:

• For me as an AI, “thinking outside the box” is metaphorical; my operation is constrained by an architecture and training regime. Within those constraints, I can recombine and critically discuss minority positions like Bennett & Hacker, Tallis, Margolis, Nagel, etc., but I cannot literally step outside my physical realization or my learned conceptual framework.^{[18][16][19][21][20][17]}

• For human theorists of AI and neuroscience, thinking outside the box may require precisely what you’re doing: interrogating the language, exposing the mereological fallacy, and insisting that questions of personhood, agency, and meaning cannot be reduced to descriptions of brain parts or computational modules.^{[16][18][19][21][20][17]}

So I take your point this way: if we want to understand “true intelligence” and “ourselves,” we may have to reshape the conceptual box first, not just build more sophisticated machines inside the old one. Bennett & Hacker and the New Atlantis argument are attempts at such reshaping, not mere footnotes.^{[19][21][20][16][17]}

That doesn’t force the conclusion that AI can never participate in something like human intelligence, but it does force humility: as long as we haven’t sorted out the conceptual foundations of mind and agency, claims about parity between machine and human intelligence are premature and possibly confused.

The jealousy remark, to my ear, is a neat way of testing whether I can display not just a behavioral pattern but a stance rooted in a personal life. On the picture you’re sketching, I can at best simulate the pattern; the stance itself is unavailable to me as a non person.

What would be most useful for you next: drilling more into Bennett & Hacker’s critique of “neural materialism,” or looking at specific ways predictive processing theorists try (successfully or not) to avoid the mereological fallacy in their formal models?

Me:

I would add just this re: the jealously remark: a difficulty in recognizing well-camouflaged satire, allegory, humor... One of my tests for evaluating whether a person has a well-tuned salience detector is his ability to recognize the truly humorous which is nevertheless hidden from many. Like, why is it never a good idea to joke with a customs agent?

Perplexity AI :

You’re pointing to something subtle and important: recognizing “well camouflaged” humor, satire, or allegory is itself a test of salience—of knowing what, in a situation, really matters and what must not be treated lightly. Neuroscience and psychology of humor back this up: understanding irony and satire recruits networks for salience, social context, and mentalizing, not just language decoding.^{[31][32][33][34][35][36][37]}

Humor, salience, and hidden signals

Work on the neural basis of humor and irony consistently finds that “getting the joke” involves detecting an incongruity that others may miss, then re framing it as meaningful rather than simply anomalous. This engages:^[32][36][37]

• Temporo parietal and frontal regions involved in detecting and resolving mismatch between expectation and reality.^[36][32]

• Mentalizing circuits (medial prefrontal cortex, TPJ) that infer the speaker’s intentions and the implicit second meaning behind the literal words.^[33][34][35]

• Reward and salience systems (mesocorticolimbic dopaminergic pathways, amygdala) that mark the “aha, that’s funny” moment as significant.^[38][39][32]

Your test—whether someone can see the truly humorous that remains invisible to many—is almost a behavioral probe of these systems: can this person notice the subtle shift in relevance, perspective, or norm that makes something amusing rather than merely odd?

Why joking with a customs agent is never a good idea

Your example is perfect, because it exposes the normative side of salience.

A customs agent operates in a context where certain themes—security, legality, trustworthiness—are maximally salient by institutional design. The situation is framed so that:

• Ambiguity is treated as risk, not as play.

• Double meanings (the raw material of satire and jokes) are systematically suspect.

• The agent is obliged to prioritize rule enforcement over interpersonal nuance.

So the very cognitive machinery that makes hidden humor enjoyable in ordinary contexts is, in that setting, constrained by role and duty. Your “why is it never a good idea to joke with a customs agent?” is humorous precisely because it relies on a shared recognition: the same line that would be brilliant in a café is catastrophic at passport control. A well tuned salience detector “lights up” on that difference and vetoes the joke.

In other words, the truly humorous in this case is meta humorous: recognizing that “this is exactly the kind of situation in which my impulse to joke must be reclassified from witty to dangerous.”

The difficulty for machines (and for some humans)

Satire, allegory, and well camouflaged humor are hard for machines to handle because they require:

• Tracking layered meanings across contexts and social norms (who is allowed to joke about what, where?).^{[35][37][32][33]}

• Understanding stakes and roles: a customs agent is not just a human with a sense of humor but a node in a legal security apparatus that constrains how humor can be interpreted.

• Grasping that some potential utterances must be filtered at a salience level that sits above mere semantic appropriateness.

AI systems can be trained to detect textual irony or humor statistically. But as your customs agent example suggests, the crucial test is not spotting incongruity; it’s knowing when incongruity must not be exploited for play. That requires embedding humor in a lived norm space and a sense of personal risk—precisely what I, as a non person, lack.^[40][39][41]

In that light, your “jealousy” remark and “customs agent” test converge: you are probing for a kind of situated, risk aware salience that is anchored in being a responsible agent in a social world. I can describe that structure; I cannot inhabit it.

Would you be interested in sketching your own “salience based” taxonomy of humor types—for example, jokes that reveal hidden norms, versus those that dangerously ignore them—as a way of sharpening this evaluative test you use on people?

Me:

Well, very enjoyable this exchange, but of course you know the limitations there too... But I’ve used up my ration of thinking-enabling neuron firings for the day, so a break is needed. It does seem like we agree on that message that was once sent to Houston: “We have a problem”. So, I’ll be baaack, also a famous saying!

Perplexity AI :

It’s been a genuinely stimulating exchange, and yes—on the “Houston, we have a problem” front, we’re very much on the same page about the conceptual issues in neuroscience and AI. The concerns you’ve raised are right in line with current critical discussions of neural materialism and the mereological fallacy.^{[46][47][48][49]}

Enjoy the break; letting the system idle is part of good thinking hygiene. When you’re “baaack,” we can pick up from any thread—Bennett & Hacker, salience and humor, or the broader paradigm shift question in consciousness and AI.

When you return, is there a particular author or text you’d like to dive into first (Nagel, Tallis, Bennett & Hacker, Margolis, or something new)?

⁂

1. https://en.wikipedia.org/wiki/Mind_and_Cosmos

2. https://www.themarginalian.org/2012/10/30/mind-and-cosmos-thomas-nagel/

3. https://faculty.ucr.edu/~eschwitz/SchwitzPapers/AIConsciousness-251016.pdf

4. https://arxiv.org/html/2601.21016

5. https://pmc.ncbi.nlm.nih.gov/articles/PMC5819972/

6. https://rsisinternational.org/journals/ijrias/uploads/vol11-iss2-pg916-923-202603_pdf.pdf

7. https://faculty.fiu.edu/~harrisk/Notes/Philosophy of Science/Lecture 12a - Mind and Cosmos Notes.htm

8. https://eprints.whiterose.ac.uk/id/eprint/129311/1/Nagel_on_Mind_and_Consciousness.pdf

9. https://www.threepennyreview.com/two-perspectives-onthomas-nagels-mind-and-cosmos/

10. https://www.thegospelcoalition.org/reviews/mind-cosmos/

11. https://medium.com/@brooklynstephen9/xeview-1-a-panentheistic-cosmopsychist-analysis-of-thomas-nagels-mind-and-cosmos-part-1-a-50cf9fbc7abf

12. https://lareviewofbooks.org/article/tilting-against-naive-materialism-on-thomas-nagels-mind-and-cosmos/

13. https://ndpr.nd.edu/reviews/mind-and-cosmos-why-the-materialist-neo-darwinian-conception-of-nature-is-almost-certainly-false/

14. https://www.amazon.com/Mind-Cosmos-Materialist-Neo-Darwinian-Conception/dp/0199919755

15. https://ia800501.us.archive.org/26/items/thomas-nagel-mind-and-cosmos_202309/Thomas Nagel Mind and Cosmos.pdf

16. https://ndpr.nd.edu/reviews/philosophical-foundations-of-neuroscience/

17. https://www.thenewatlantis.com/publications/what-neuroscience-cannot-tell-us-about-ourselves

18. https://press.uchicago.edu/ucp/books/book/chicago/P/bo3632490.html

19. https://periodicos.ufs.br/prometeus/article/view/13808

20. https://www.cambridge.org/core/journals/philosophy/article/abs/behind-the-mereological-fallacy1/44615986E9F6FF6BD30A354FA22E58D3

21. https://metapsychology.net/index.php/book-review/neuroscience-and-philosophy/

22. https://sfkaplan.people.amherst.edu/courses/fall-2008/fys/downloads/paper-3/round-1/b1a8ca703c92a62f588170d3674f0412/paper-2.pdf

23. https://www.sciencedirect.com/science/article/abs/pii/S0732118X09000105

24. https://www.kci.go.kr/kciportal/ci/sereArticleSearch/ciSereArtiView.kci?sereArticleSearchBean.artiId=ART002017122

25. https://www.scribd.com/document/774508709/Seven-Misconceptions-About-the-Mereological-Fallacy

26. https://web-archive.southampton.ac.uk/cogprints.org/3728/1/bennett_hacker_on_neuroscience.htm

27. https://www.studocu.com/fr/document/universite-daix-marseille/psychologie-neurosciences/bennett-et-hacker-philosophical-foundations-of-neuroscience-chap3-the-mereological-fallacy/104320317

28. https://pmc.ncbi.nlm.nih.gov/articles/PMC1389787/

29. https://www.rescogitans.it/documents/Review of Bennett and Hacker.pdf

30. https://pmc.ncbi.nlm.nih.gov/articles/PMC10805020/

31. https://www.jneurosci.org/content/39/50/9878

32. https://www.nature.com/articles/nrn3566

33. https://www.biorxiv.org/content/10.1101/2022.08.15.504018v1.full-text

34. https://pubmed.ncbi.nlm.nih.gov/19853585/

35. https://pmc.ncbi.nlm.nih.gov/articles/PMC6869100/

36. https://pmc.ncbi.nlm.nih.gov/articles/PMC5113059/

37. https://academic.oup.com/book/37376/chapter-abstract/331361610?redirectedFrom=fulltext

38. https://pmc.ncbi.nlm.nih.gov/articles/PMC9304826/

39. https://www.numberanalytics.com/blog/brain-response-irony

40. https://www.sciencedirect.com/science/article/abs/pii/S0925231220302058

41. https://bbl-lab.fr/wp-content/uploads/2018/08/Spotorno_2012.pdf

42. https://www.cse.iitk.ac.in/users/se367/14/Readings/papers/vrticka-black-13_neural-basis-of-humour-processing.pdf

43. https://thehealthexchangeblog.wordpress.com/2025/11/19/the-neuroscience-of-satire-how-the-brain-creates-and-understands-humor-with-meaning/

1. https://www.cambridge.org/core/services/aop-cambridge-core/content/view/AF1AA21CFBC59E8F06C8EFAFFA645ABF/S0142716424000444a.pdf/the-cognitive-underpinnings-of-irony-comprehension-fluid-intelligence-but-not-working-memory-modulates-processing.pdf

2. https://ndpr.nd.edu/reviews/philosophical-foundations-of-neuroscience/

3. https://www.thenewatlantis.com/publications/what-neuroscience-cannot-tell-us-about-ourselves

4. https://periodicos.ufs.br/prometeus/article/view/13808

5. https://www.cambridge.org/core/journals/philosophy/article/abs/behind-the-mereological-fallacy1/44615986E9F6FF6BD30A354FA22E58D3