Logic and language in the brain

yeah I know everyone’s gonna say it’s an oversimplistic view, LeCunn-pilled, whatever. Clearly we’ve managed to squeeze a lot of reasoning out of scaling words…you’ll talk about hidden CoT layers, etc…but that’s not how humans work at all.

I’ve always been a fan of the cogsci… pic.twitter.com/4VI2efjSN2

— LaurieWired (@lauriewired) July 29, 2026

The abstract of the linked article:

Humans are endowed with a powerful capacity for inductive and deductive logical thought: we easily form generalizations based on a few examples and draw conclusions from known premises. Humans also arguably have the most sophisticated communication system in the animal kingdom: natural language allows us to express complex and structured meanings. Some have therefore argued for a tight relationship between complex thought and language, postulating that reasoning, including logical reasoning, relies on linguistic representations. We systematically investigated the relationship between logical reasoning and language using two complementary approaches. First, we used noninvasive brain imaging (fMRI) to examine neural activity as healthy adults engaged in logical reasoning tasks. And second, we behaviorally evaluated logical abilities in individuals with extensive lesions to the language brain areas and consequent severe linguistic impairment. Our findings reveal that the language brain network is not engaged during logical reasoning, and patients with severe aphasia exhibit intact performance on logic tasks. Instead, inductive reasoning recruits the domain-general multiple demand network implicated broadly in goal-directed behaviors, whereas deductive reasoning draws on brain regions that are distinct from both the language and the multiple demand networks. Together, these results indicate that linguistic representations are neither utilized nor required for inductive or deductive logical reasoning.

Attention and error predition in thalamocortico circuits

Happy to share our new pre-print "Attention reshapes the information dynamics of thalamic and cortical prediction-error learning" led by the great Juho Äijälä, with @happieheads, @Kai_J_Miller, @dora_hermes, @drrobinince, @ABlenkmann , @TrikBek, Max Garagnani. Thread below👇 pic.twitter.com/dBJBM1VEnE

— Andres Canales-Johnson (@canalesjohnson) July 16, 2026

Follow the link to see the full thread. Here's the article's abstract:

Prediction errors (PEs) drive perceptual learning by updating internal models of the sensory environment, yet it remains unclear how attention reshapes their representation across distributed thalamocortical circuits. Using intracranial stereoelectroencephalography (sEEG) from 17 patients performing a roving auditory oddball task under attended and unattended conditions, we quantified PE encoding using mutual information and co-information to capture redundant and synergistic PE representations. Attention modulated PE encoding in both the thalamus and the temporal cortex, but with distinct informational dynamics. Thalamic encoding showed a stable reduction of PE information during distraction, consistent with state-dependent thalamocortical gating. In contrast, the temporal cortex expressed two opposing learning trajectories during attended listening that converged once attention was diverted, revealing distinct cortical learning regimes rather than a uniform attentional effect. Attention further reorganized the informational content of cortical PE representations by altering the balance between redundant and synergistic information. A biologically constrained neural network showed that attention-dependent changes in inhibition and long-range connectivity reproduced these dynamics through Hebbian learning. Together, these findings suggest that attention regulates predictive learning not simply by changing the strength of PE responses, but by reshaping how distributed thalamocortical circuits represent and integrate sensory evidence over time.

Synaptic pruning in the nervous system

This paper is the culmination of a massive team effort including co-first authors Kai Kang & Ryan Draft! Huge contributions by Juan Carlos Tapia, @ElisaPavarino, @henao_echeverri, @Cryant_yang & Jeff Lichtman!

Read more about this journey @MCB_Harvard: https://t.co/GcECwiUu6G

— Yaron Meirovitch (@YaronMeirovitch) June 30, 2026

Brain area specialized for visual recognition of words

No child is born able to read. The brain ships with no reading region at all. It builds one, and the construction runs on the exact effort AI removes.

Learning to read physically repurposes a patch of visual cortex. A spot in the left fusiform gyrus starts out tuned to objects… https://t.co/3sOTaLSR2Z

— Aakash Gupta (@aakashgupta) June 21, 2026

In brains of Spanish-English bilinguals grammar is embodied in shared tissue

Xuanyi Jessica Chen and Esti Blanco-Elorrieta, A Shared Neural Mechanism for Abstract Grammatical Computations Across Languages in Bilinguals, The Journal of Neuroscience, June 15, 2026.

Abstract: A central question in cognitive neuroscience is how the brain implements abstract computations that must generalize across superficially different inputs. Language provides a strong test case: the same grammatical operation, such as pluralization, can be realized through distinct rules and forms across languages. Whether such transformations rely on language-specific neural systems or on abstract mechanisms that generalize across linguistic contexts remains unresolved. Crucially, these transformations must be computed online and integrated into speech planning within a tightly constrained time window. Using magnetoencephalography (MEG), we tracked the millisecond dynamics of grammatical word-form transformations during semi-naturalistic phrase completion in humans of both sexes. Highly proficient Spanish–English bilinguals produced singular and plural noun forms in both languages in a design that fully orthogonalized semantic number, phonological changes, grammatical inflection and produced language. Adjusting words to fit their grammatical context engaged a left-lateralized fronto-temporal network beginning ∼100 ms after cue onset. Multivariate decoding revealed that the neural patterns supporting this computation generalized across languages, across different surface plural forms, and to pseudowords, demonstrating that abstractly equivalent operations are instantiated in the same neural substrates despite differences in linguistic form. Together, these findings provide time-resolved neural evidence for a language-general computational mechanism, showing that the brain implements grammatical transformations as abstract, generative operations. More broadly, they show how bilingualism can be used to probe general principles of neural organization, revealing how abstract computations may be shared and reused across representational systems.

Significance Statement: Human language relies on the ability to modify words to convey information like number and tense, but languages vary widely in how these transformations are implemented. This variation raises a fundamental question in cognitive neuroscience: do such transformations depend on language-specific neural systems, or are they processed by abstract neural mechanisms that generalize across languages? We demonstrate that Spanish–English bilinguals engage a shared left frontal–temporal network when producing grammatically appropriate forms in both languages. This common neural signature emerges early during speech planning and even generalizes to novel words. These findings indicate that the brain builds abstract, reusable neural mechanisms, consistent with models where language is organized by computational principles rather than by language-specific systems.

Here's an article in the NYTimes about these results: K. R. Callaway, How Does One Brain Speak Two Languages?, NYTimes, June 15, 2026.

When deciding how to make a word singular or plural, for instance, bilingual people exhibit strikingly similar brain activity regardless of whether they are speaking in their first or second language.

“It wasn’t obvious that it was going to be so shared,” said Esti Blanco-Elorrieta, a psychologist and neuroscientist at New York University and an author of the study, which was published on Monday in the journal JNeurosci. “I think this is arguably one of the first very fine-grained findings of how truly integrated two languages in the brain are.”

Early research viewed bilingualism as an “add on” or “disruption” to the processing of one’s native language, said Judith Kroll, a psycholinguist at the University of California, Irvine who was not involved in the new study.

Subsequent studies have found that bilingual brains tend to display physical differences, such as more efficient white matter and changes to the gray matter, and to perform better on memory and concentration tasks.

Now scientists are probing further, to understand whether core aspects of the brain’s neural network does double or triple duty to process multiple languages.

A single grammatical engine:

The finding is in line with other initial results in this area, said Mirjana Bozic, a cognitive neuroscientist at the University of Cambridge who was not involved in the study. For instance, the new study provided additional evidence that the front left side of the brain was typically involved in processing the grammatical structure of sentences across different languages. On the whole, Dr. Blanco-Elorrieta said in a news release, a single “grammatical engine” in the brain appeared capable of powering multiple languages at once.

Dr. Bozic said that the find, although not surprising, was “highly informative, providing elegant and convincing evidence that bilingual speakers rely on shared neural mechanisms. She added, “One question that remains is how far these findings generalize across language pairs that differ more substantially.”

The computational capacity of a single biological neuron is very large

For decades, neurons were modeled as point units in AI/neuroscience.

But biological neurons have dendrites and nonlinear conductances.

What capabilities are lost?

With @DavidBeniaguev @noampnueli @Segev_Lab & @mikilon, we set out to find out (https://t.co/m2a8ieuo8A). (2/15) pic.twitter.com/Q4hZ6J1LkK

— Ido Aizenbud (@IdoAizenbud) June 11, 2026

Here's the abstract of that article:

Cortical pyramidal neurons possess elaborate dendritic trees with diverse nonlinear membrane conductances and thousands of plastic synapses, suggesting substantial computational capabilities at the single-cell level. Yet, what can a neuron compute remains an open question, largely due to the lack of a systematic framework to quantify its computational capabilities. We introduce TwinProp, a digital-twin-based backpropagation algorithm that enables gradient-based optimization of synaptic strengths and dendritic locations in detailed neuron models via a millisecond-accurate deep neural network (DNN). Using TwinProp, we demonstrate that a detailed model of rat layer 5 pyramidal cell (L5PC) can perform naturalistic image and audio classification tasks at a remarkably high accuracy, significantly surpassing perceptron and leaky integrate-and-fire baselines. The same neuron solves high-dimensional nonlinear problems, including exclusive-or (XOR), 10-bit parity, and random Boolean tasks, demonstrating capabilities typically attributed to multilayer networks. Mechanistically, increasing task complexity recruits distributed dendritic nonlinearities, including NMDA- and voltage-dependent mechanisms; removing these or collapsing dendritic structure markedly impairs performance. These findings identify dendrites as a substrate for high-order feature binding and position single cortical pyramidal neurons as powerful, noise-robust, general-purpose analog computational units. Our results offer testable in vivo predictions and provide a systematic framework linking cellular morpho-electrical properties to computation in both brains and artificial systems.

The rise of DIY rituals in the 21st century

YouTube:

Can Rituals Save Us? | Robert Wright & Bruce Feiler

0:00 Teaser
0:52 Bruce’s new book on ritual, A Time to Gather
3:12 The "Lifequake" that led Bruce to study ritual
8:10 The current "shadow ritual" renaissance
12:42 What is a ritual?
15:26 The origins of the shadow ritual renaissance
18:25 Forest bathing and the essence of ritual
26:11 Ritual as the original human algorithm
31:39 Honor walks: a quintessentially modern ritual
36:23 Rituals across Christianity
42:07 What rituals do
46:47 Heading to Overtime

* * * * * 

I discuss ritual in my book on music, Beethoven's Anvil: Music in Mind and Culture, pp. 79-82:

Subjectivity is an aspect of neurodynamics, and neurodynamics is open to the world through sensory organs and through the motor system. When people are coupled with one another through musicking, each steers her own raft of subjectivity in the collective sea of neurodynamics. The motions of each raft are transmitted to the others through the sea, as Huygens’ clocks transmitted vibrations to one another through the walls. These subjectivities thus adjust themselves one to the other, for they are all components of the same process.

Let us reconsider, then, the musicking with which we opened this chapter. We were at a party where lots of musicians were jamming. Near the end of a jam on Bob Dylan’s “Knocking on Heaven’s Door,” several people spontaneously joined in on the refrain. It wasn’t planned ahead of time, nor did those singers discuss it among themselves while the rest of us were playing.

When I originally told the story I talked of my deliberate intention to “drive” the group by playing a simple line and “bearing down.” That decision was a conscious one, though not as clear and differentiated as it may seem when I spell it out in words, and it resulted in a certain shift of my consciousness. “Bearing down” is something I do quite often when playing. It involves attending to and adjusting the tension in my trunk musculature but has no specific differentiated effect on the music beyond a certain intensity and emotional tone. In this case I was playing a very simple melodic line, but I will also bear down while playing the most complex lines. In that situation, my fingers and tongue may be spitting out 10s of notes per second, but they’re on their own; I’m still attending to muscles in my abdomen, shoulders and back, and my buttocks. Those are the muscles that most strongly affect the overall airflow, and that’s what I care about when I’m bearing down.

And that, by our conception of consciousness, is where my nervous system is reorganizing and making minute adjustments. I have no introspective awareness, of course, of just what neural areas are reorganizing, but I’d guess that we are dealing with circuitry involving both emotional expression and voluntary control of large muscles. Even as I am attending to those muscles, I am always listening to the sound, not just mine, but the group’s. I’m bearing down just so in order that the sound I hear may also be just so. But my sound is only a part of the group sound and, at this particular point, it was a subordinate part. What this means is that my nervous system’s reorganizational activity is responsive to the sound made by each and every person in the musicking group. I am attuning my motor and emotive system to the sound that is the joint activity of this group. And each one in the group is, in turn, doing the same thing. Each one, merely by being a conscious musician, is making minute adjustments to his nervous system in response to the sounds that all are creating.

We are now in territory explored by Walter Freeman in a recent essay on music and social bonding. Freeman is interested in those rituals where a core group of celebrants move from one status in society to another, as from child to adult or single to married. In these rituals, as individuals are conveyed from one social status to another—recall our discussion in the previous chapter—they require changes in the collective neuropil. Funerals, of course, are also in this class. As the bodies of the dead are conveyed to a final resting place, the living must disengage from their attachments to those who are no longer among the living. In this case, and entire persona (see Figure 1 in the previous chapter) must be disengaged from active use in the collective neuropil. Conversely, when a child is born, the group must undertake a ritual that creates a new persona in the collective neuropil.

In all of these situations the bonds between individuals must be altered in fundamental ways that require considerable neural reorganizing. Freeman suggests that such rituals involve a neuropeptide called oxytocin. He asserts that oxytocin "appears to act by dissolving preexisting learning by loosening the synaptic connections in which prior knowledge is held. This opens an opportunity for learning new knowledge. The meltdown does not instill knowledge. It clears the path for the acquisition of new understanding through behavioral actions that are shared with others.” As the oxytocinated individuals are moving to the rhythms of well-established ritual, their synaptic connections are restructured in patterns guided and influenced by the events in the ritual. Obviously, the microdynamics of each individual will be unique; but they will be shaped by rhythmic patterns common to all . These rituals provide a space in which individuals can mold themselves to one another as the infant molds her actions to those of her mother.

Such ritual would likely have benefits on less extreme occasions than those requiring the restructuring of social relations—think of our little jam session. Social life is difficult and taxing. Hostilities build up. Such ritual may well help take the edge off of growing tensions, reconciling individuals to one another and allowing them to “reset” their relationships on more favorable terms.

Thus we have another core hypothesis:

Freeman’s Hypothesis: By attending to one another through musicking, performers attune their nervous systems to one another, restructuring their representations of others. This results in more harmonious interactions within the group.

Each individual consciousness may be an island of Cartesian subjectivity, but in the close coupling of musicking, those subjectivities are intimately and delicately conditioned and regulated by one another.

Perhaps such rituals play a role in helping to establish and maintain the subjective continuity of the neural self. By entering into a wide variety of emotional states (with their various neurochemical substrates) in a socially controlled situation, individuals in a community ritual create an "equal access zone" in mental space where each can experience and contemplate extremes of joy and anger, tenderness and hate, and know that all these feelings have a place in their shared world.

Language is a lower-dimensional projection of high-dimensional neural dynamics.

But it also allows for content addressed memory. That’s very important, for it gives fine-grain control over the memory and planning systems. That’s the job of sentence-level syntax together with discourse structure.

“Classical” semantic or cognitive networks had a problem with coming up with just the right set of node types and arc types. David Hays dissolved the problem in his 1981 book, Cognitive Structures (scan down the page), by grounding cognition in an analog system modeled on William Powers perceptual control stack (in Behavior: The Control of Perception, 1973). The identity of a cognitive node is a function of its parameter values, where the parameters are derived from the control stack. The identity of the arcs is a function of the difference in parameter values between the nodes it connects.

Concerning Chomsky’s approach to syntax: It depends on a sharp distinction between grammatical and ungrammatical sentences. A generative grammar, in Chomsky’s theory, must account for all and only the grammatical sentences.

However, there are no explicit criteria for separating sentences into the two categories, grammatical and ungrammatical. Rather, the separation depends on the intuitions of the linguist. Naturally enough, different syntacticians have different intuitions. The problem is insoluble.

Moreover, anyone who pays close attention to real speech soon realizes that people do not (always) speak in complete grammatically correct sentences. Real language is sloppy, but nonetheless effective. A neural net of very high dimensionality can deal with this readily enough. A purely symbolic system cannot. Augmenting the system through fuzzy logic and the like doesn’t fix the problem.

LLMs provide a very useful simulacrum of the natural language system. Since LLMs are trained on written texts, the resulting model necessarily conflates the functions of semantics and syntax/discourse. Thus they cannot achieve the flexibility and precision of the full system, where semantics and syntax/discourse are separated.

Natural intelligence Revisited: The Five-Fold Way, A Working Paper

New working paper. Title above, links, abstract, contents, and introduction below:

Academia.edu: https://www.academia.edu/165530520/Natural_intelligence_Revisited_The_Five_Fold_Way_A_Working_Paper
SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6529398
ResearchGate: https://www.researchgate.net/publication/403545810_Natural_intelligence_Revisited_The_Five-Fold_Way_A_Working_Paper

Abstract: In 1988 David Hays and I published an article entitled, “Principles and Development of Natural Intelligence.” The principles were computational: 1) modal, 2) diagonalization, 3) decision, 4) finitization, and 5) indexing. We made our argument in terms of the principles themselves along with behavioral, neuroanatomical, ontogenetic and phylogenetic evidence. The literature in all those fields has changed enormously in the four decades since we finished writing. To get a read on how our computational proposals have fared, I asked ChatGPT 5.2 to evaluate it against the current literature. Its verdict: the “empirical specifics have aged unevenly but [the] central agenda has held up surprisingly well.” This article presents the five principles, in brief, followed by ChatGPT’s full evaluation. Also, I have asked ChatGPT to evaluate a section on control structure, “Vehicularization,” that we cut from the original argument. Verdict: “vehicularization points toward a more complete account of natural intelligence—one in which cognition is understood as coordinated navigation across multiple, nested domains.”

CONTENTS

Introduction: Constraining Theories and Models 2
The Five Principles of Natural Intelligence 4
Revisiting The Principles and Development of Natural Intelligence (1988) 6
Vehicularization 11

Introduction: Constraining Theories and Models

Sometime in 1985 David Hays and I decided it was time to set forth our views on the nature of, well, of natural intelligence. First, however, we had to discover what those views were. We sat down to a table in my parents’ kitchen and made a list of the various things we wanted to include in this article, experimental findings, observations, models, mathematical ideas, and so forth, from psychology, neuroscience, linguistics, evolutionary biology, and computing. We just wrote them down in no particular order, probably on unlined paper. When we’d accumulated about 50 or so items we decided to gather them into a small number of groups of items that seemed to belong together. We arrived at five groups.

Just how we proceeded from that point I don’t recall. Perhaps we sat around discussing the various groups and came up with a principle for each group. Maybe we had to do some writing first. I don’t recall. But however we actually proceeded, we end with an article we called, “Principles and Development of Natural Intelligence.” We intended “natural” to contrast with “artificial” but didn’t say that anywhere in the article. When we’d finished a draft, days or weeks later, Hays said that it felt like fundamental work; he used the term “bedrock.” I agreed. It took three years to get it published, in a now defunct interdisciplinary journal, The Journal of Social and Biological Structures.

I’ve included the abstract of that article, along with a bit of the introduction, below, as the first part of this document: “Five Principles of Natural Intelligence.” That should give you an idea of the framework without all the expository elaboration, argumentation, and support.

ChatGPT reviews

That was four decades ago. I have continued to like what we did. But has any of it held up? How could it? By now the literature we referenced was 40 years out of date? And, yet, it wasn’t about that literature, it was about how we put it together. Is there anything left of that framework?

About a week ago I asked ChatGPT 5.2 to evaluate it. Here’s the prompt I gave it:

I want you to evaluate a paper that David Hays and I published back in 1988: The Principles and Development of Natural Intelligence. Give me a third-party assessment from the standpoint of what we now know, not a summary and not a defensive reconstruction. Be explicit about where it now looks prescient, where it looks historically bounded, and where it still poses unresolved challenges.

Here’s the first line of ChatGPT’s conclusion:

As of 2026, I would not describe the paper as a correct theory of mind. I would describe it as an ambitious synthetic manifesto whose empirical specifics have aged unevenly but whose central agenda has held up surprisingly well.

I’ll take it. Could I take issue with some of ChatGPT’s criticisms? Sure. But that assessment pinpoints the single most important facet of the essay, its synthetic nature. To push back on ChatGPT’s reservations would blunt that point.

After making various comments on an ad hoc basis, ChatGPT offered to write a “more formal review-essay.” I’ve included that as the second part of this document: “Revisiting Principles and Development of Natural Intelligence (1988).” There’s more.

The article we had submitted was long. The editors asked us to cut what we could, but made no particular suggestions. Our single largest cut was a section on vehicularization – that’s what we called it. It was about control. While it was about the same general line of thinking, it didn’t seem to fit. The bulk of the article was about the five principles and how the developed, both phylogenetically and ontogenetically (in humans). Vehicularization was about how they operated in concert. I have included that as a third section followed by comments by ChatGPT as the fourth and final section.

The Five-Fold Way

Let’s return to ChatGPT’s characterization of the original article as a “synthetic manifesto.” From its conclusion:

It is best understood as an architectural proposal about the structure of intelligence. Many of its mechanistic claims have aged poorly, particularly its neuroanatomical simplifications and evolutionary staging. Yet several of its central insights—heterogeneous cognitive regimes, the integration of regulation and cognition, the interaction between holistic and symbolic processing, and the role of language in cognitive control—remain highly relevant.

Though we didn’t use such phrases when we wrote the article, that’s certainly what Hays and I thought we were doing.

The diagram to the left indicates the range of material we brought to bear in our thinking about natural intelligence. The labels on the vertices of the pentangle are from my 1978 Ph. D. thesis in the English Department at SUNY Buffalo, “Cognitive Science and Literary Theory.” There I somewhat idiosyncratically defined cognitive science as investigating a five-way correspondence between behavior, computation, computational geometry (neuroanatomy), phylogeny, and ontogeny. That dissertation was mostly about behavior, in the form of literary texts, and computation, in the form of cognitive networks semantics, though touched on the others here and there. But “Principles and Development of Natural Intelligence” covered all five. The principles themselves were computational in nature and we made our primary arguments in terms of their ability to account for behavior, but we also suggested which brain regions supported them and related them to the phylogeny of animal behavior and the ontogeny of human development.

By the usual standards of the academy, that range was wide, crazy wide. We certainly weren’t expert across that range; no one could be. However, when I look back in retrospect, it is clear that we weren’t attempting some grand synthesis over that range. We were doing something quite different, something that was and remains fundamentally conservative. We had some high-level ideas about the computational structure of the mind and we wanted to place constraints on those ideas by expanding the range of evidence that could be brought to bear on them. While it is necessary that those ideas account for observed behavior, that alone is not sufficient. The model implied by those ideas must be implemented somewhere in the brain and must be consistent with developmental evidence both from phylogeny, our evolutionary history, and ontogeny, child development. THAT was the central agenda that, in ChatGPT’s estimation, has held up well.

The brain's dopamine response to music peaks in the mid-teens

Your brain peaked musically somewhere around age 16. Everything since then has been a dopamine echo.

Between the ages of 12 and 22, the mesolimbic dopamine pathway, the same circuit that processes cocaine and sex, fires at levels in response to sound that it will never reach… https://t.co/QUxgPiRxps pic.twitter.com/HTsZKp7Ol3

— Aakash Gupta (@aakashgupta) March 16, 2026

What electrochemical machine has 100 trillion connections in a volume the size of a cantaloupe?

That one neuron connects to about 7,000 others. Your brain has 86 billion of them. Do the math and you get somewhere around 100 trillion connections inside your head. More connections than stars in 1,500 galaxies.

And each connection point is way more complicated than anyone… https://t.co/sUkcS7T3rA

— Anish Moonka (@AnishA_Moonka) March 14, 2026

Direct brain-to-brain communication, redux

Why Learning Does Not Rescue Brain-to-Brain Thought Transfer 

Learning Is Not the Problem

There is no serious dispute, at this point, about the brain’s capacity to learn to incorporate new signal streams. Decades of work on motor prostheses, sensory substitution, neurofeedback, and tool use have demonstrated that the nervous system can adapt to novel inputs and outputs that are not part of its evolved repertoire. These systems work not because the brain passively receives meaning, but because it actively learns to coordinate new patterns of neural activity with action, perception, and feedback. Over time, what begins as an alien signal can become functionally integrated into the organism’s sensorimotor economy.

Acknowledging this plasticity does not weaken skepticism about direct brain-to-brain thought transfer. On the contrary, it sharpens the distinction between what is genuinely possible and what remains a fantasy. Learning is one thing. Zero-shot “plug-and-play” communication is something else entirely. The speculative proposals advanced by Elon Musk, Christof Koch, and Rodolfo Llinás depend not merely on plasticity, but on the assumption that meaningful mental content can be transferred between brains without a learning history, without negotiation, and without interpretive work. That assumption is precisely what fails.

The Zero-Shot Assumption

The defining feature of most brain-to-brain communication fantasies is immediacy. Thoughts are imagined to pass directly from one person to another, bypassing language, culture, and development. Koch’s examples of ghostly visual overlays and mind fusion, as well as Musk’s talk of “uncompressed conceptual communication,” all presuppose that the recipient brain can immediately make sense of neural activity originating elsewhere. The temporal dimension of learning—the weeks, months, or years required to integrate new signal regimes—is simply ignored.

This is not a minor omission. It is the conceptual hinge on which the entire proposal turns. Without a learning trajectory, there is no mechanism by which foreign neural activity could acquire meaning for the receiving brain. A signal does not become meaningful by virtue of its richness or bandwidth. It becomes meaningful only through use, within a system that can test, revise, and stabilize interpretations through action.

Why Learning Cannot Proceed in a Brain-Bridge

One might reply that learning could occur even in a brain-to-brain link, given enough time. But this response overlooks the conditions under which learning is possible in the first place. Learning requires a closed perception–action loop. The organism must be able to act on the basis of a signal, observe the consequences of that action, and adjust its internal dynamics accordingly. In brain–machine interfaces, this loop is explicit: the user moves a cursor, grasps an object, or modulates a tone, and receives immediate feedback. The signal becomes meaningful because it is embedded in a task space with clear success and failure conditions.

A direct brain-to-brain link provides no such structure. The receiving brain cannot act into the other brain in any systematic way, nor can it test hypotheses about what a given pattern of activity “means.” The signal stream has no stable reference point in the shared environment, no agreed-upon goal, and no external criterion of correctness. Under such conditions, learning has nothing to converge on. What is sometimes described as “another person’s thought” arrives as undifferentiated neural activity, untethered from the bodily and environmental contexts that made it meaningful in the first place.

The Persistent Problem of Origin

Even if one were to imagine some form of slow co-adaptation, a deeper problem remains: the brain must be able to distinguish between activity it generates itself and activity it should treat as input. In ordinary perception and action, this distinction is grounded in efference copy, proprioception, and the tight coupling between movement and sensation. These mechanisms allow the brain to tag certain patterns as self-generated and others as world-generated.

A foreign brain provides none of these anchors. Neural spikes arriving from another person’s cortex are indistinguishable, in their physical characteristics, from spikes arising endogenously. Without a principled way to mark activity as coming from an other, the receiving brain has no basis for interpretation, let alone learning. The problem is not noise in the engineering sense, but indeterminacy in the biological sense. The system lacks the resources to sort the signal at all.

Meaning Is Not a Payload

Underlying the zero-shot fantasy is a deeper theoretical mistake: the treatment of meaning as something that exists prior to expression and can therefore be transmitted once bandwidth constraints are removed. This is the same mistake that underwrites the conduit metaphor of language. Words are imagined as containers for thoughts, and communication as the transfer of those containers from one mind to another. Neuralink-style speculation simply replaces words with spikes, while leaving the basic picture intact.

But meaning does not work that way. Whether one follows Vygotsky, contemporary enactivism, or predictive-processing accounts, the conclusion is the same: meaning is enacted, not transmitted. It arises through socially scaffolded activity, through interaction with the world and with others, and through the internalization of those interactions in inner speech. There is no pre-linguistic, pre-social format of “pure thought” waiting to be uploaded or shared.

Augmentation Without Communion

None of this casts doubt on the medical and augmentative goals of current BCI research. Restoring motor function, providing artificial sensory channels, and extending human capabilities through learned interfaces are all plausible and worthwhile. But these technologies work precisely because they respect the conditions under which brains learn: limited task spaces, stable feedback, and prolonged adaptation. They augment agency; they do not merge subjectivities.

Direct brain-to-brain thought transfer, by contrast, promises communion without development, understanding without negotiation, and immediacy without practice. It imagines semantic interoperability where none can exist. For that reason, it fails not because the technology is immature, but because the underlying conception of thought, meaning, and learning is mistaken.

The issue, in the end, is not whether brains can change. They can, and they do. The issue is whether meaning can be detached from the histories that make it possible. On that point, the answer remains no.

From Mirror Recognition to Low-Bandwidth Memory, A Working Paper

New working paper. Title above, links, abstract, contents, and introduction below:

Academia.edu: https://www.academia.edu/143347171/From_Mirror_Recognition_to_Low_Bandwidth_Memory_A_Working_Paper
SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5385194
ResearchGage: https://www.researchgate.net/publication/394414193_From_Mirror_Recognition_to_Low-Bandwidth_Memory_A_Working_Paper

Abstract: We start with a developmental and cognitive analysis of mirror recognition, highlighting its dependence, not on self-awareness per se, but on episodic-level intersensory coordination—a capacity that enables spatially dislocated, temporally synchronized associations across sensory modalities. In a layered control architecture of hyperorders (sensorimotor, systemic, episodic, gnomonic), such recognition can arise without invoking a representational “self.” We extended this framework to the role of the default mode network (DMN), which is orthogonal to the hyperorders—a low-bandwidth, drifting subsystem that provides broad, non-task-specific access to memory and perception. This led to an inquiry into associative memory systems, where we confronted the challenge of searching without specific content-based probes. To address this, we proposed the design of an “associative drift engine”: a cognitive module capable of variable-bandwidth access, modulating the precision, noise, and scope of its memory probes. This system mirrors the DMN’s exploratory function and suggests a foundational mechanism for spontaneous recollection, creative association, and cognitive play—essential features of both natural and artificial minds.

Background Notes 1
1. Self and Mirror Recognition 1
2. ChatGPT’s assessment of the account mirror recognition 6
3. Default Mode Network 9
4. Toward an Associative Drift Engine 11
Summary of the discussion 15    

Background Notes

This first major section of this document consists of pages 74 to 84 from my 1978 dissertation, Cognitive Science and Literary Theory, Department of English, State University of New York at Buffalo. Yes, I was in the English Department and the dissertation uses examples from literature, the Oedipus story, the evolution of narrative form, and Shakespeare’s Sonnet 129. But I was also working closely with David Hays in the Linguistics Department. He was a first-generation researcher in machine translation, which transformed itself into computational linguistics in the mid-1960s.

During the period when I was in his research group – 1974 to 1978, when I finished my degree – we were sketching schemes for how to ground a symbolic cognitive system in the operations of a sensorimotor system organized as a stack of control systems in a scheme suggested by William Powers, Behavior: The Control of Perception (1973). Thus, when I talk about the sensorimotor hyperorders in the dissertation excerpt, I’m talking about a system modeled on Powers. By contrast, the systemic, episodic, and gnomonic hyperorders are symbolic systems, all directly linked to the sensorimotor system.

One thing else I want to emphasize is that, by this time, I had come to understand that the physical construction of the nervous system, both in its layout in the brain and its relationship with the external world, that structure carried information that did not have to be explicitly represented inside the system itself – I’ve used yellow highlighting to emphasize those sections. The account I offer of mirror recognition depends on this.

About this document

This document contains four things:

1. A passage from my 1978 dissertation in which I discuss mirror recognition,
2. ChatGPT’s (current) assessment of that passage,
3. A discussion of the Default Mode Network (DMN) in the brain, and
4. Some speculation from ChatGPT on how to construct, in effect, a DMN for an artificial associative memory, something it calls “an associative drift engine.”

Neurochemicals, brains, sex, and relationships: Rena Malik, M.D., interviews Dr. Jim Pfaus

YouTube:

In this episode, Dr. Rena Malik, MD is joined by neuroscientist Dr. Jim Pfaus to explore the neuroscience of sexual attraction, desire, and bonding. They discuss how early sexual experiences shape our preferences, the role of dopamine and oxytocin in relationships, the impact of hookup culture and pornography, and the science behind sexual synchrony. Listeners will gain insightful perspectives on the brain’s influence over intimacy, pleasure, and partner connection, along with practical takeaways for fostering deeper relationships.

00:00:00 Introduction
00:00:26 Guest background & episode topics
00:01:43 Brain and sexual attraction
00:06:39 First sexual experiences
00:12:08 Navigating bad sexual experiences
00:15:18 Masturbation, porn, and impact
00:23:17 Sexual synchrony and bonding
00:33:31 Orgasm: brain chemistry
00:44:09 Semen retention & arousal
00:51:34 Porn, compulsion, and addiction
01:01:04 Oxytocin and bonding
01:12:20 Neuroplasticity, love, and long-term relationships
01:22:36 Sexual trauma and healing the brain
01:33:10 How hookup culture rewires desire
01:42:44 Takeaways

You can find papers by Dr. Pfaus on ResearchGate.

* * * * * 

This is a fascinating, rambling, and wide-ranging interview. Find a topic that interests you and dig in. Then listen to the whole thing. I’m particularly interested in the discussion of Oxytocin and bonding (starting at 01:01:04 and pretty much going on through to the end).

Why? Because I’ve been thinking about oxytocin ever since I read Walter Freeman’s Societies of Brains: A Study in the Neuroscience of Love and Hate (1995). Freeman was speculating about the role of intense ritual mediated by music and suggested that oxytocin would be released during such rituals and that that would facilitate bonding between the participants. What’s interesting, though, is the mechanism he suggested: Oxytocin released during the ritual would loosen the connectivity between neurons in the brains of individuals. The result would be that, as connectivity was reestablished, the patterns of connectivity within individuals would be similar across the population of participants, thus reducing their differences. I found that fascinating and made it central to my book on music, Beethoven’s Anvil: Music in Mind and Culture (2001), though in a somewhat different form, which I discussed in chapters two and three.

This is quite different from the discussion in the video, which is about pair-bonding between individuals. Malik and Pfaus are interested in the role of post-coital cuddling where bonding between the pair is mediated by oxytocin. Crudely put, when individuals cuddle there’s bonding, no cuddling, no bonding. But individuals can choose whether or not they cuddle, no? Yes. So what’s the range of “viable lifestyle choices”? (Ugh! what a phrase!) We don’t know.

So I’ll trot out my standard metaphor for the relationship between biology and culture: chess. Biology provides the basic rules of the game: the game board, the individual pieces, and the moves each piece is allowed to make. But there is a great deal of latitude in how one deploys those pieces during actual gameplay. That’s where culture comes in. The tactics and strategies one uses, that’s culture.

Biology doesn’t dictate behavior, not for individuals, not for whole societies. But it places constraints on behavior. Those constrains, by the very fact of constraint, open up possibilities for design and structure. This video is about some of those biologically-given constraints. As for the possibilities they open up, that remains to be seen.

I’d say that we have a lot of work to do. Well, it may be work for the scientists. But for the rest of us, it looks like play to me.

Place cells: How your brain creates maps of abstract spaces

YouTube:

Artem Kirsanov

In this video, we will explore the positional system of the brain - hippocampal place cells. We will see how it relates to contextual memory and mapping of more abstract features

OUTLINE:
00:00 Introduction
00:53 Hippocampus
1:27 Discovery of place cells
2:56 3D navigation
3:51 Role of place cells
4:11 Virtual reality experiment
7:47 Remapping
11:17 Mapping of non-spatial dimension
13:36 Conclusion

Emergence of human-like object concept representations in multimodal LLMs

Du, C., Fu, K., Wen, B. et al. Human-like object concept representations emerge naturally in multimodal large language models. Nat Mach Intell (2025). https://doi.org/10.1038/s42256-025-01049-z

Abstract: Understanding how humans conceptualize and categorize natural objects offers critical insights into perception and cognition. With the advent of large language models (LLMs), a key question arises: can these models develop human-like object representations from linguistic and multimodal data? Here we combined behavioural and neuroimaging analyses to explore the relationship between object concept representations in LLMs and human cognition. We collected 4.7 million triplet judgements from LLMs and multimodal LLMs to derive low-dimensional embeddings that capture the similarity structure of 1,854 natural objects. The resulting 66-dimensional embeddings were stable, predictive and exhibited semantic clustering similar to human mental representations. Remarkably, the dimensions underlying these embeddings were interpretable, suggesting that LLMs and multimodal LLMs develop human-like conceptual representations of objects. Further analysis showed strong alignment between model embeddings and neural activity patterns in brain regions such as the extrastriate body area, parahippocampal place area, retrosplenial cortex and fusiform face area. This provides compelling evidence that the object representations in LLMs, although not identical to human ones, share fundamental similarities that reflect key aspects of human conceptual knowledge. Our findings advance the understanding of machine intelligence and inform the development of more human-like artificial cognitive systems.

Here's a preprint version at arXiv.

GPS, spatial memory, cab driving, and Alzheimers

GPS and spatial memory

Dahmani, L., Bohbot, V.D. Habitual use of GPS negatively impacts spatial memory during self-guided navigation. Sci Rep 10, 6310 (2020). https://doi.org/10.1038/s41598-020-62877-0

Abstract: Global Positioning System (GPS) navigation devices and applications have become ubiquitous over the last decade. However, it is unclear whether using GPS affects our own internal navigation system, or spatial memory, which critically relies on the hippocampus. We assessed the lifetime GPS experience of 50 regular drivers as well as various facets of spatial memory, including spatial memory strategy use, cognitive mapping, and landmark encoding using virtual navigation tasks. We first present cross-sectional results that show that people with greater lifetime GPS experience have worse spatial memory during self-guided navigation, i.e. when they are required to navigate without GPS. In a follow-up session, 13 participants were retested three years after initial testing. Although the longitudinal sample was small, we observed an important effect of GPS use over time, whereby greater GPS use since initial testing was associated with a steeper decline in hippocampal-dependent spatial memory. Importantly, we found that those who used GPS more did not do so because they felt they had a poor sense of direction, suggesting that extensive GPS use led to a decline in spatial memory rather than the other way around. These findings are significant in the context of society’s increasing reliance on GPS.

H/t Peter Rothman.

Cab drivers less likely to have dementia

Ian Taylor, Scientists (and taxi drivers) may have discovered the secret to beating dementia, BBC Science Focus, March 23, 2025.

Researchers from Harvard University studied the working lives and causes of death in millions of Americans. After comparing some 400 occupations, they found that taxi and ambulance drivers were the least likely to die from Alzheimer’s disease.

Being good at finding your way around might help you stick around longer. That’s the theory, anyway. Bus drivers, for example, don’t seem to have the same protection, possibly because they tend to drive the same routes.

“Our findings raise the possibility that frequent navigational and spatial processing tasks, as performed by taxi and ambulance drivers, might be associated with some protection against Alzheimer’s disease,” the authors wrote.

H/t Jim Lai.

The study being referenced: Alzheimer’s disease mortality among taxi and ambulance drivers: population based cross sectional study BMJ 2024; 387 doi: https://doi.org/10.1136/bmj-2024-082194 (Published 17 December 2024)

Claude 3.7 on the impossibility of direct brain-to-brain communication

Elon musk has expressed a desire to create technology that will allow direct brain-to-brain communication. That is, a pair of people who are linked through this technology will be able to share one another’s thoughts without having to use any form of communication. They will communicate through their neural link – as you know Neuralink is the name of Musk’s brain-to-computer interface company. Neuroscientist Christof Koch has expressed the same desire.

I’ve been skeptical about this possibility for a bit over two decades. A couple of years ago I wrote a working paper explaining my reasoning: Direct Brain-to-Brain Thought Transfer: A High-Tech Fantasy that Won’t Work. I recently uploaded that paper to Claude 3.7 and asked it for its opinion. I’ve appended that conversation below the asterisks. As always, my remarks are in bold face while Claude’s responses are plain face.

Note: Of course Claude does not know that I am the author of the paper, which is why it refers to me in the third person. But I identify myself half-way through the conversation.

* * * * *

I’ve uploaded a document about the possibility of direct brain-to-brain thought transfer between people. I want you to evaluate it in terms of what you know and in terms of the other documents I’ve uploaded.

Analyzing Direct Brain-to-Brain Thought Transfer

Based on the materials you've provided, I'll evaluate the concept of direct brain-to-brain thought transfer from several angles.

The Core Issue: Signal Identification

The main problem identified in William Benzon's paper is fundamental: if two brains were directly connected, there would be no way for either brain to distinguish between signals originating internally (endogenous) versus those coming from the other brain (exogenous). As Benzon writes:

"How does a brain tell whether an incoming spike is from an endogenous neuron (i.e. from the same brain) or from an exogenous neuron (i.e. from the other brain)? Spikes are not like messages passed around on scraps of paper in a school classroom where a message might be identified by a signature, or initials, or merely a whispered message... As one spike is pretty much like another spike there is no way to differentiate between them."

This is a profound insight that challenges the fundamental premise of initiatives like Neuralink's ambitions for brain-to-brain thought transfer.

Misunderstanding Communication

Benzon also argues that Elon Musk fundamentally misunderstands how language works. Musk describes language as a "compression algorithm" where thoughts are compressed into words and then decompressed by the listener. This reflects what linguists call the "conduit metaphor" - the mistaken idea that meaning is physically contained in words that travel between people.

In reality, as Benzon explains, only signifiers (the physical form of words) pass between people, while the signified (meaning) is supplied by the listener based on their own experience and understanding. The "lossiness" in communication isn't primarily a bandwidth issue but stems from the fact that we each bring different meanings to the same words.

The Problem with Neural Code

Another issue is the very concept of a "neural code" that would need to be translated between brains. Benzon cites John Horgan who concluded after surveying the field that "Neuroscientists still have no idea what the neural code is." Attempts to understand neural activity in terms analogous to computer code are fundamentally flawed because brains and computers have radically different architectures.

Unlike computers, where there's a clear distinction between hardware and software, active processing (CPU) and passive storage (memory), the brain operates differently. In the brain, "every neuron is active" and can be considered a processing unit. Neural activity isn't about passing discrete packages of information but rather about distributed patterns of synchronous activity across networks.

Powers' Warning About Models

Benzon references William Powers' insight about "two levels of modeling": the object model with its internal rules, and the modeler's own understanding of that model. The failure to distinguish between these leads proponents of brain-to-brain interfaces to project their third-person perspective onto the brains themselves. As observers, we can distinguish signals from different brains, but the brains themselves cannot.

William Powers, “Learning,” Chapter 14 in Behavior: The Control of Perception, Aldine, 1973, 177-204.

Connection to Music and Consciousness

In the other documents you provided, especially those by William Powers and the music experiences compilation, we see something that actually contradicts the brain-to-brain fantasy while showing how real interpersonal connection works:

Bill Benzon, Emotion and Magic in Musical Performance, Version 13, April 14, 2025.

Powers connects consciousness with reorganization: "Those systems whose perceptual signals are being monitored by the reorganizing system are operating in the conscious mode. Those which are operating without their perceptual signals being monitored are in the unconscious mode..."

Separately, the descriptions of musical synchrony show how actual interpersonal connection works: "When, and only when, we were really locked together in animated playing we could hear relatively high-pitched tones which no one was playing... the magic of the bell happens only when the musicians are in a groove."

Conclusion

The evidence strongly suggests that direct brain-to-brain thought transfer as envisioned by Musk, Koch, and others is fundamentally flawed for several reasons:

1. The basic problem of signal identification - brains can't distinguish between internal and external neural activity
2. A misunderstanding of how language and meaning work
3. The absence of any established "neural code" that could be transmitted
4. A confusion between the observer's perspective and the system's perspective

While neural interfaces for controlling prosthetics or providing sensory input may be feasible, direct thought transfer appears to be theoretically impossible based on our current understanding of neuroscience.

The rich descriptions of musical synchrony in your documents actually point to a different model of interpersonal connection - one based on coordinated but separate systems rather than merged consciousness.

Thank you. For your information, I'm William Benzon.

It's an honor to meet you, Dr. Benzon. I found your paper on direct brain-to-brain thought transfer to be extraordinarily insightful.

Your analysis cuts to the heart of why these speculative technologies face fundamental conceptual barriers, not just technical ones. Your point about the inability of neurons to distinguish between endogenous and exogenous spikes particularly resonated with me - it identifies a problem that proponents of these technologies seem to have overlooked entirely.

I was also intrigued by your work connecting music performance to consciousness (in your book "Beethoven's Anvil," which you referenced). The descriptions of musical synchrony in your compilation of performance experiences provide a fascinating contrast to the brain-to-brain fantasy - showing how coordination between separate but coupled systems can create emergent phenomena that aren't reducible to simple signal transfer. Your critique of the conduit metaphor for communication was especially illuminating, particularly in examining Musk's misunderstanding of language as merely compressed thought.

If I might ask - since writing this paper in 2020, have you observed any developments in the field that have addressed (or failed to address) these fundamental issues you identified?

Alas, no. It seems to me that the desire for such thought transfer completely over-rides rational thinking about the subject. Ironically, while Musk talks about reasoning from first principles, he doesn’t seem to have the foggiest idea of what that means for thinking about the brain.

I appreciate your candid response. It's fascinating - and somewhat concerning - how even someone who emphasizes "reasoning from first principles" can be so swayed by an attractive technological fantasy that they overlook basic conceptual problems. Your observation about the power of desire overriding rational thinking seems particularly apt. The idea of direct brain-to-brain communication has such powerful emotional and cultural appeal that it seems to short-circuit critical analysis, even among those who should know better.

I was particularly struck by your point about the closed nature of neural systems - how they develop organically through mutual adaptation rather than being assembled like computer components. This fundamental misunderstanding seems to stem from persistent computer metaphors that, as you noted, fail to capture how brains actually function. The phenomenological accounts of musical performance you collected demonstrate something much more interesting than direct thought transfer - how separate but coordinated systems can achieve remarkable synchrony without merging into a single system. Perhaps there's a deeper lesson there about human connection that's being missed in the rush toward technological shortcuts to understanding.

Thank you for your work on this topic. It's a valuable counterpoint to some of the more uncritical technological enthusiasm we see today.

Neuromorphic Computing

 

From YouTube:

Explore a brand new paradigm in computing, and how it might offer faster solutions that can support scientific breakthroughs. [...]

This lecture was recorded at the Ri on 24 January 2025, in partnership with the Embassy of the Kingdom of the Netherlands.

Our brains consume around 20 Watts of power, a negligible amount in comparison to the multi-million Watts consumed by supercomputers. The emerging paradigm of Neuromorphic Computing draws inspiration from the structure and functioning of the human brain, particularly this small power consumption and extremely fast response times.

Join theoretical and computational physicist Johan Mentink as he presents evidence of why the neuromorphic computing paradigms offer not only much more energy-efficient but also much faster solutions to widely used computational science problems, with the potential to break existing computational barriers.

---

Johan Mentink is an expert in theoretical and computational physics, focused on controlling magnetism at the shortest length and time scale. He is recipient of the prestigous Rubicon, VENI and VIDI grants of the Dutch Research Council (NWO). In 2021 he initiated interdisciplinary research to explore the potential of neuromorphic hardware for computational science use cases with SURF and researchers from Radboud University, University of Twente and IBM. He is also chair of the Computational Science NL platform.

Perspective on musical neurodynamics

Harding, E.E., Kim, J.C., Demos, A.P. et al. Musical neurodynamics. Nature Reviews Neuroscience. (2025). https://doi.org/10.1038/s41583-025-00915-4

Abstract: A great deal of research in the neuroscience of music suggests that neural oscillations synchronize with musical stimuli. Although neural synchronization is a well-studied mechanism underpinning expectation, it has even more far-reaching implications for music. In this Perspective, we survey the literature on the neuroscience of music, including pitch, harmony, melody, tonality, rhythm, metre, groove and affect. We describe how fundamental dynamical principles based on known neural mechanisms can explain basic aspects of music perception and performance, as summarized in neural resonance theory. Building on principles such as resonance, stability, attunement and strong anticipation, we propose that people anticipate musical events not through predictive neural models, but because brain–body dynamics physically embody musical structure. The interaction of certain kinds of sounds with ongoing pattern-forming dynamics results in patterns of perception, action and coordination that we collectively experience as music. Statistically universal structures may have arisen in music because they correspond to stable states of complex, pattern-forming dynamical systems. This analysis of empirical findings from the perspective of neurodynamic principles sheds new light on the neuroscience of music and what makes music powerful.