Biology subjects

Piantadosi, S. T.

Publications and source records attributed to Piantadosi, S. T..

8 recordsLinked to original sources

bioRxiv2025

A human brain network specialized for abstract formal reasoning

Humans stand out in the animal kingdom for their ability to reason in highly abstract ways. Using a deep-data precision fMRI approach, we identify and richly characterize a network of frontal brain areas that support abstract formal reasoning. This abstract reasoning network robustly dissociates from the domain-general Multiple Demand network--the current leading candidate substrate of fluid intelligence--as well as from three other networks supporting high-level cognition: the language network, the intuitive physical reasoning network, and the social reasoning network. Finally, the areas of this network respond robustly during both deductive and inductive reasoning, during classic matrix reasoning problems, and when solving multiplication and division problems. This network may therefore support the most abstract forms of reasoning, possibly constituting a human-specific adaptation.

neuroscience↗

bioRxiv2025

Evidence from Formal Logical Reasoning Reveals that the Language of Thought is not Natural Language

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 non-invasive 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. SignificanceWhich cognitive mechanisms allow humans to reason logically, to understand whether a conclusion follows from the premises? Are they the same ones that allow the assembly of words into structured representations? Scholars have debated for millennia whether logical reasoning is inextricably tied to natural language, or instead relies on a distinct "language of thought" (LOT). Using fMRI in healthy adults and evaluating logical ability in individuals with severe aphasia, we find that distinct neural systems support language processing vs. logical (inductive and deductive) reasoning. These results suggest that, at least in mature brains, language processing does not underpin logical inference, perhaps due to the distinct representational format of the logical LOT.

neuroscience↗

bioRxiv2025

Attention is all you need (in the brain): semantic contextualization in human hippocampus

Word meanings in language are contextualized by surrounding words. Inspired by the self-attention mechanism in transformer-based large language models (LLMs), we hypothesized that structural composition in the brain arises from combining canonical (non-contextual) word representations with those of nearby words. We analyzed single unit activity in the human hippocampus, a region involved in semantic and contextual processing, while n=10 participants listened to podcasts. We found that hippocampal neurons encoded word position within a clause, using both ordinal and frequency-domain positional encoding. Moreover, neural responses to specific words reflected both the words own lexical semantics and a weighted sum of the embeddings of preceding words. The relative weighting of these contextualizing words correlated with LLM self-attention weights. These findings suggest that contextualization in the brain makes use of vectorial shifts that have a resemblance to attentional reweighting in LLMs, and highlight the role of the mesial temporal lobe within the broader language network.

neuroscience↗

bioRxiv2025

A vectorial code for semantics in human hippocampus

As we listen to speech, our brains track the meanings of the words we hear. Recent successes of large language models suggest that distributed population geometry can capture rich semantic relationships between words. Motivated by this idea, we hypothesized that semantic information in the brain may likewise be expressed in distributed patterns of activity across neurons, rather than in the activity of neurons narrowly tuned to a specific word. We recorded responses of hundreds of neurons in the human hippocampus while participants listened to narrative speech. We find encoding of contextual word meaning in the simultaneous activity of neurons whose individual selectivities span multiple unrelated semantic categories. Decoding and population geometry analyses revealed distinct neural coding principles for low-versus high-frequency words, likely reflecting the greater polysemy of common words. Similar to embedding vectors in semantic language models, distance between neural population responses correlates with semantic distance; however, this effect was only observed in contextual embedding models (GPT-2 and BERT), suggesting that the semantic distance effect depends critically on contextualization. Consistent with this, we find that neural population activity supports a multidimensional semantic subspace that aligns most closely with the contextual structure captured by GPT-2. Moreover, for semantically similar words, even contextual embedders showed an inverse correlation between semantic and neural distances; we attribute this pattern to the noise-mitigating benefits of contrastive coding. Ultimately, these results provide a neurocomputational account for understanding how neural populations track word meaning.

neuroscience↗

bioRxiv2018

High local mutual information drives the response in the human language network

The fronto-temporal language network responds robustly and selectively to sentences. But the features of linguistic input that drive this response and the computations these language areas support remain debated. Two key features of sentences are typically confounded in natural linguistic input: words in sentences a) are semantically and syntactically combinable into phrase- and clause-level meanings, and b) occur in an order licensed by the languages grammar. Inspired by recent psycholinguistic work establishing that language processing is robust to word order violations, we hypothesized that the core linguistic computation is composition, and, thus, can take place even when the word order violates the grammatical constraints of the language. This hypothesis predicts that a linguistic string should elicit a sentence-level response in the language network as long as the words in that string can enter into dependency relationships as in typical sentences. We tested this prediction across two fMRI experiments (total N=47) by introducing a varying number of local word swaps into naturalistic sentences, leading to progressively less syntactically well-formed strings. Critically, local dependency relationships were preserved because combinable words remained close to each other. As predicted, word order degradation did not decrease the magnitude of the BOLD response in the language network, except when combinable words were so far apart that composition among nearby words was highly unlikely. This finding demonstrates that composition is robust to word order violations, and that the language regions respond as strongly as they do to naturalistic linguistic input as long as composition can take place.

neuroscience↗

bioRxiv2018

Learning list concepts through program induction

Humans master complex systems of interrelated concepts like mathematics and natural language. Previous work suggests learning these systems relies on iteratively and directly revising a language-like conceptual representation. We introduce and assess a novel concept learning paradigm called Marthas Magical Machines that captures complex relationships between concepts. We model human concept learning in this paradigm as a search in the space of term rewriting systems, previously developed as an abstract model of computation. Our model accurately predicts that participants learn some transformations more easily than others and that they learn harder concepts more easily using a bootstrapping curriculum focused on their compositional parts. Our results suggest that term rewriting systems may be a useful model of human conceptual representations.

animal behavior and cognition↗

bioRxiv2018

Monkeys predict trajectories of virtual prey using basic variables from Newtonian physics.

The demands of foraging are a major driver in the evolution of cognitive faculties. To successfully pursue a mobile prey that is attempting to avoid capture, the ability to predict its flight path can provide a crucial advantage. We hypothesized that, during pursuit, rhesus macaques exploit patterns in preys behavior to predict the preys future positions. We modeled behavior of three macaques in a joystick-controlled pursuit task in which prey follow simple escape algorithms that involve repulsion from the subject and from the walls of the virtual enclosure. We find that, even in this artificial task, macaques actively predict and aim towards preys future positions, increasing their foraging success. Their predictions are derived from the three core variables in Newtonian dynamics: position, velocity, and acceleration. Even after extensive training, subjects favored these principles and ignored other regularities in prey behavior. Most notably, they ignored the effects their own actions would have on the prey, despite extensive training and even though doing so would have further improved performance. We conjecture that subjects have a strong bias towards using physical principles to pursue fleeing prey, possibly reflecting an evolved physics module. The observed predictive behavior suggests that foraging demands facilitate the development of prospection.

animal behavior and cognition↗

bioRxiv2016

Robust mixture modeling reveals category-free selectivity in reward region neuronal ensembles

Classification of neurons into clusters based on their response properties is an important tool for gaining insight into neural computations. However, it remains unclear to what extent neurons fall naturally into discrete functional categories. We developed a Bayesian method that models the tuning properties of neural populations as a mixture of multiple types of task-relevant response patterns. We applied this method to data from several cortical and striatal regions in economic choice tasks. In all cases, neurons fell into only two clusters: one mixed-selectivity cluster containing all task-sensitive cells and another of no selectivity (i.e. pure noise) cells. The single cluster of task-sensitive cells argues against robust categorical tuning in these areas. The no selectivity cells were unanticipated; their identification allows for improved measurement of ensemble effects. Our findings provide a valuable tool for analysis of neural data and place strong constraints on neurocomputational models of choice and control.

neuroscience↗