Efficient cognition requires flexible interactions between distributed neural networks in the human brain. These networks adapt to challenges by flexibly recruiting different regions and connections. In this talk, I will discuss how we study functional network plasticity and reorganization with combined neurostimulation and neuroimaging across the adult life span. I will argue that short-term plasticity enables flexible adaptation to challenges, via functional reorganization. My key hypothesis is that disruption of higher-level cognitive functions such as language can be compensated for by the recruitment of domain-general networks in our brain. Examples from healthy young brains illustrate how neurostimulation can be used to temporarily interfere with efficient processing, probing short-term network plasticity at the systems level. Examples from people with dyslexia help to better understand network disorders in the language domain and outline the potential of facilitatory neurostimulation for treatment. I will also discuss examples from aging brains where plasticity helps to compensate for loss of function. Finally, examples from lesioned brains after stroke provide insight into the brain’s potential for long-term reorganization and recovery of function. Collectively, these results challenge the view of a modular organization of the human brain and argue for a flexible redistribution of function via systems plasticity.

This webinar brings together three leaders in theoretical and computational neuroscience—Larry Abbott, Haim Sompolinsky, and Terry Sejnowski—to discuss how neural circuits generate fundamental aspects of the mind. Abbott illustrates mechanisms in electric fish that differentiate self-generated electric signals from external sensory cues, showing how predictive plasticity and two-stage signal cancellation mediate a sense of self. Sompolinsky explores attractor networks, revealing how discrete and continuous attractors can stabilize activity patterns, enable working memory, and incorporate chaotic dynamics underlying spontaneous behaviors. He further highlights the concept of object manifolds in high-level sensory representations and raises open questions on integrating connectomics with theoretical frameworks. Sejnowski bridges these motifs with modern artificial intelligence, demonstrating how large-scale neural networks capture language structures through distributed representations that parallel biological coding. Together, their presentations emphasize the synergy between empirical data, computational modeling, and connectomics in explaining the neural basis of cognition—offering insights into perception, memory, language, and the emergence of mind-like processes.

This webinar convened researchers at the intersection of Artificial Intelligence and Neuroscience to investigate how large language models (LLMs) can serve as valuable “model organisms” for understanding human language processing. Presenters showcased evidence that brain recordings (fMRI, MEG, ECoG) acquired while participants read or listened to unconstrained speech can be predicted by representations extracted from state-of-the-art text- and speech-based LLMs. In particular, text-based LLMs tend to align better with higher-level language regions, capturing more semantic aspects, while speech-based LLMs excel at explaining early auditory cortical responses. However, purely low-level features can drive part of these alignments, complicating interpretations. New methods, including perturbation analyses, highlight which linguistic variables matter for each cortical area and time scale. Further, “brain tuning” of LLMs—fine-tuning on measured neural signals—can improve semantic representations and downstream language tasks. Despite open questions about interpretability and exact neural mechanisms, these results demonstrate that LLMs provide a promising framework for probing the computations underlying human language comprehension and production at multiple spatiotemporal scales.

Dr. Nicholas Blauch will present on his work "Topoformer: Brain-like topographic organization in transformer language models through spatial querying and reweighting". Dr. Blauch is a postdoctoral fellow in the Harvard Vision Lab advised by Talia Konkle and George Alvarez. Paper link: https://openreview.net/pdf?id=3pLMzgoZSA Trends in NeuroAI is a reading group hosted by the MedARC Neuroimaging & AI lab (https://medarc.ai/fmri | https://groups.google.com/g/medarc-fmri).

Developing agents capable of modeling complex environments and human behaviors within them is a key goal of artificial intelligence research. Progress towards this goal has exciting potential for applications in video games, from new tools that empower game developers to realize new creative visions, to enabling new kinds of immersive player experiences. This talk focuses on recent advances of my team at Microsoft Research towards scalable machine learning architectures that effectively capture human gameplay data. In the first part of my talk, I will focus on diffusion models as generative models of human behavior. Previously shown to have impressive image generation capabilities, I present insights that unlock applications to imitation learning for sequential decision making. In the second part of my talk, I discuss a recent project taking ideas from language modeling to build a generative sequence model of an Xbox game.

Developing agents capable of modeling complex environments and human behaviors within them is a key goal of artificial intelligence research. Progress towards this goal has exciting potential for applications in video games, from new tools that empower game developers to realize new creative visions, to enabling new kinds of immersive player experiences. This talk focuses on recent advances of my team at Microsoft Research towards scalable machine learning architectures that effectively capture human gameplay data. In the first part of my talk, I will focus on diffusion models as generative models of human behavior. Previously shown to have impressive image generation capabilities, I present insights that unlock applications to imitation learning for sequential decision making. In the second part of my talk, I discuss a recent project taking ideas from language modeling to build a generative sequence model of an Xbox game.

Executive functions are cognitive processes that allow us to plan, monitor and execute our goals. Using fMRI, we investigated how early deafness influences crossmodal plasticity and the organisation of executive functions in the adult human brain. Results from a range of visual executive function tasks (working memory, task switching, planning, inhibition) show that deaf individuals specifically recruit superior temporal “auditory” regions during task switching. Neural activity in auditory regions predicts behavioural performance during task switching in deaf individuals, highlighting the functional relevance of the observed cortical reorganisation. Furthermore, language grammatical skills were correlated with the level of activation and functional connectivity of fronto-parietal networks. Together, these findings show the interplay between sensory and language experience in the organisation of executive processing in the brain.

Recent insights from auditory neuroscience provide a new perspective on how the brain encodes speech. Using these recent insights, I will provide an overview of key factors underpinning individual differences in children’s development of language and phonology, providing a context for exploring atypical reading development (dyslexia). Children with dyslexia are relatively insensitive to acoustic cues related to speech rhythm patterns. This lack of rhythmic sensitivity is related to the atypical neural encoding of rhythm patterns in speech by the brain. I will describe our recent data from infants as well as children, demonstrating developmental continuity in the key neural variables.

We are now clearly entering a new age in our relationship with machines. The power of AI natural language processors and image generators has rapidly exceeded the expectations of even those who developed them. Serious questions are now being asked about the extent to which machines could become — or perhaps already are — sentient or conscious. Do AI machines understand the instructions they are given and the answers they provide? In this talk I will consider the prospects for conscious machines, by which I mean machines that have feelings, know about their own existence, and about ours. I will suggest that the recent focus on information processing in models of consciousness, in which the brain is treated as a kind of digital computer, have mislead us about the nature of consciousness and how it is produced in biological systems. Treating the brain as an energy processing system is more likely to yield answers to these fundamental questions and help us understand how and when machines might become minds.

Young children have sophisticated representations of their visual and linguistic environment. Where do these representations come from? How much knowledge arises through generic learning mechanisms applied to sensory data, and how much requires more substantive (possibly innate) inductive biases? We examine these questions by training neural networks solely on longitudinal data collected from a single child (Sullivan et al., 2020), consisting of egocentric video and audio streams. Our principal findings are as follows: 1) Based on visual only training, neural networks can acquire high-level visual features that are broadly useful across categorization and segmentation tasks. 2) Based on language only training, networks can acquire meaningful clusters of words and sentence-level syntactic sensitivity. 3) Based on paired visual and language training, networks can acquire word-referent mappings from tens of noisy examples and align their multi-modal conceptual systems. Taken together, our results show how sophisticated visual and linguistic representations can arise through data-driven learning applied to one child’s first-person experience.

The ability to build, store, and manipulate semantic representations lies at the core of all our (inter)actions. Combining evidence from cognitive neuroimaging and experimental neuropsychology, I study the neurocognitive correlates of semantic knowledge in relation to other cognitive functions, chiefly language. In this talk, I will start by reviewing neuroimaging findings supporting the idea that semantic representations are encoded in distributed yet specialized cortical areas (1), and rapidly recovered (2) according to the requirement of the task at hand (3). I will then focus on studies conducted in neurodegenerative patients, offering a unique window on the key role played by a structurally and functionally heterogeneous piece of cortex: the anterior temporal lobe (4,5). I will present pathological, neuroimaging, cognitive, and behavioral data illustrating how damages to language-related networks can affect or spare semantic knowledge as well as possible paths to functional compensation (6,7). Time permitting, we will discuss the neurocognitive dissociation between nouns and verbs (8) and how verb production is differentially impacted by specific language impairments (9).

Metaphor is a pervasive phenomenon in language and cognition. To date, the vast majority of psycholinguistic research on metaphor has focused on noun-noun metaphors of the form An X is a Y (e.g., My job is a jail). Yet there is evidence that verb metaphor (e.g., I sailed through my exams) is more common. Despite this, comparatively little work has examined how verb metaphors are processed. In this talk, I will propose a novel account for verb metaphor comprehension: verb metaphors are understood in the same way that analogies are—as comparisons processed via structure-mapping. I will discuss the predictions that arise from applying the analogical framework to verb metaphor and present a series of experiments showing that verb metaphoric extension is consistent with those predictions.

Socially Assistive Robots (SARs) have shown great potential to help children in therapeutic and healthcare contexts. SARs have been used for companionship, learning enhancement, social and communication skills rehabilitation for children with special needs (e.g., autism), and mood improvement. Robots can be used as novel tools to assess and rehabilitate children’s communication skills and mental well-being by providing affordable and accessible therapeutic and mental health services. In this talk, I will present the various studies I have conducted during my PhD and at the Cambridge Affective Intelligence and Robotics Lab to explore how robots can help assess and rehabilitate children’s communication skills and mental well-being. More specifically, I will provide both quantitative and qualitative results and findings from (i) an exploratory study with children with autism and global developmental disorders to investigate the use of intelligent personal assistants in therapy; (ii) an empirical study involving children with and without language disorders interacting with a physical robot, a virtual agent, and a human counterpart to assess their linguistic skills; (iii) an 8-week longitudinal study involving children with autism and language disorders who interacted either with a physical or a virtual robot to rehabilitate their linguistic skills; and (iv) an empirical study to aid the assessment of mental well-being in children. These findings can inform and help the child-robot interaction community design and develop new adaptive robots to help assess and rehabilitate linguistic skills and mental well-being in children.

Learning verbs is difficult and critical to learning one's native language. Children appear to benefit from seeing multiple events and comparing them to each other, and structural alignment theory provides a good theoretical framework to guide research into how preschool children may be comparing events as they learn new verbs. The talk will include 6 studies of early verb learning that make use of eye-tracking procedures as well as other behavioral (pointing) procedures, and that test key predictions from SA theory including the prediction that seeing similar examples before more varied examples helps observers learn how to compare (progressive alignment) and the prediction that when events have very low alignability with other events, that is one cue that the events should be ignored. Whether or how statistical learning may also be at work will be considered.

We will explore how neuropsychology can be leveraged to directly test cognitive neuroscience theories using the case of frontotemporal dementias affecting the language network. Specifically, we will focus on pathological, neuroimaging, and cognitive data from primary progressive aphasia. We will see how they can help us investigate the reading network, semantic knowledge organisation, and grammatical categories processing. Time permitting, the end of the talk will cover the temporal dynamics of semantic dimensions recovery and the role played by the task.

Understanding Human ability to learn novel concepts from just a few sensory experiences is a fundamental problem in cognitive neuroscience. I will describe a recent work with Ben Sorcher and Surya Ganguli (PNAS, October 2022) in which we propose a simple, biologically plausible, and mathematically tractable neural mechanism for few-shot learning of naturalistic concepts. We posit that the concepts that can be learned from few examples are defined by tightly circumscribed manifolds in the neural firing-rate space of higher-order sensory areas. Discrimination between novel concepts is performed by downstream neurons implementing ‘prototype’ decision rule, in which a test example is classified according to the nearest prototype constructed from the few training examples. We show that prototype few-shot learning achieves high few-shot learning accuracy on natural visual concepts using both macaque inferotemporal cortex representations and deep neural network (DNN) models of these representations. We develop a mathematical theory that links few-shot learning to the geometric properties of the neural concept manifolds and demonstrate its agreement with our numerical simulations across different DNNs as well as different layers. Intriguingly, we observe striking mismatches between the geometry of manifolds in intermediate stages of the primate visual pathway and in trained DNNs. Finally, we show that linguistic descriptors of visual concepts can be used to discriminate images belonging to novel concepts, without any prior visual experience of these concepts (a task known as ‘zero-shot’ learning), indicated a remarkable alignment of manifold representations of concepts in visual and language modalities. I will discuss ongoing effort to extend this work to other high level cognitive tasks.

What do people do when they comprehend language in discourse? According to many psychologists, they build and maintain cognitive representations of utterances in four complementary mental models for discourse that interact with each other: the surface text, the text base, the situation model, and the context model. When people encounter metaphors in these utterances, they need to incorporate them into each of these mental representations for the discourse. Since influential metaphor theories define metaphor as a form of (figurative) analogy, involving cross-domain mapping of a smaller or greater extent, the general expectation has been that metaphor comprehension is also based on analogizing. This expectation, however, has been partly borne out by the data, but not completely. There is no one-to-one relationship between metaphor as (conceptual) structure (analogy) and metaphor as (psychological) process (analogizing). According to Deliberate Metaphor Theory (DMT), only some metaphors are handled by analogy. Instead, most metaphors are presumably handled by lexical disambiguation. This is a hypothesis that brings together most metaphor research in a provocatively new way: it means that most metaphors are not processed metaphorically, which produces a paradox of metaphor. In this talk I will sketch out how this paradox arises and how it can be resolved by a new version of DMT, which I have described in my forthcoming book Slowing metaphor down: Updating Deliberate Metaphor Theory (currently under review). In this theory, the distinction between, but also the relation between, analogy in metaphorical structure versus analogy in metaphorical process is of central importance.

Analogical reasoning –learning about new things by relating it to previous knowledge– lies at the heart of human intelligence and creativity and forms the core of educational practice. Children start creating and using analogies early on, making incredible progress moving from associative processes to successful analogical reasoning. For example, if we ask a four-year-old “Horse belongs to stable like chicken belongs to …?” they may use association and reply “egg”, whereas older children will likely give the intended relational response “chicken coop” (or other term to refer to a chicken’s home). Interestingly, despite state-of-the-art AI-language models having superhuman encyclopedic knowledge and superior memory and computational power, our pilot studies show that these large language models often make mistakes providing associative rather than relational responses to verbal analogies. For example, when we asked four- to eight-year-olds to solve the analogy “body is to feet as tree is to …?” they responded “roots” without hesitation, but large language models tend to provide more associative responses such as “leaves”. In this study we examine the similarities and differences between children's and six large language models' (Dutch/multilingual models: RobBERT, BERT-je, M-BERT, GPT-2, M-GPT, Word2Vec and Fasttext) responses to verbal analogies extracted from an online adaptive learning environment, where >14,000 7-12 year-olds from the Netherlands solved 20 or more items from a database of 900 Dutch language verbal analogies.

Biologically plausible spiking neural networks (SNNs) are an emerging architecture for deep learning tasks due to their energy efficiency when implemented on neuromorphic hardware. However, many of the biological features are at best irrelevant and at worst counterproductive when evaluated in the context of task performance and suitability for neuromorphic hardware. In this talk, I will present an alternative paradigm to design deep learning architectures with good task performance in real-world benchmarks while maintaining all the advantages of SNNs. We do this by focusing on two main features – event-based computation and activity sparsity. Starting from the performant gated recurrent unit (GRU) deep learning architecture, we modify it to make it event-based and activity-sparse. The resulting event-based GRU (EGRU) is extremely efficient for both training and inference. At the same time, it achieves performance close to conventional deep learning architectures in challenging tasks such as language modelling, gesture recognition and sequential MNIST.

Language appears to be the most complex system of animal communication described to date. However, its precursors were present in the communication of our evolutionary ancestors and are likely shared by our modern ape cousins.  All great apes, including humans, employ a rich repertoire of vocalizations, facial expressions, and gestures. Great ape gestural repertoires are particularly elaborate, with ape species employing over 80 different gesture types intentionally: that is towards a recipient with a specific goal in mind. Intentional usage allows us to ask not only what information is encoded in ape gestures, but what do apes mean when they use them. I will discuss recent research on ape gesture, on how we approach the question of decoding meaning, and how with new methods we are starting to integrate long overlooked aspects of ape gesture such as group and individual variation, and expression and emotion into our study of these signals.

Biologically plausible spiking neural networks (SNNs) are an emerging architecture for deep learning tasks due to their energy efficiency when implemented on neuromorphic hardware. However, many of the biological features are at best irrelevant and at worst counterproductive when evaluated in the context of task performance and suitability for neuromorphic hardware. In this talk, I will present an alternative paradigm to design deep learning architectures with good task performance in real-world benchmarks while maintaining all the advantages of SNNs. We do this by focusing on two main features -- event-based computation and activity sparsity. Starting from the performant gated recurrent unit (GRU) deep learning architecture, we modify it to make it event-based and activity-sparse. The resulting event-based GRU (EGRU) is extremely efficient for both training and inference. At the same time, it achieves performance close to conventional deep learning architectures in challenging tasks such as language modelling, gesture recognition and sequential MNIST

How do children learn to read in different countries? How do deficits in various components of the reading network affect learning outcomes? What are the consequences of such deficits in different languages? In this talk, I will present a full-blown developmentally plausible computational model of reading acquisition that has been implemented in English, French, Italian and German. The model can simulate individual learning trajectories and intervention outcomes on the basis of three component skills: orthography, phonology, and vocabulary. I will use the model to show how cross-language differences affect the learning-to-read process in different languages and to investigate to what extent similar deficits will produce similar or different manifestations of dyslexia in different languages.

We examine consciousness from the perspective of theoretical computer science (TCS), a branch of mathematics concerned with understanding the underlying principles of computation and complexity, including the implications and surprising consequences of resource limitations. We propose a formal TCS model, the Conscious Turing Machine (CTM). The CTM is influenced by Alan Turing's simple yet powerful model of computation, the Turing machine (TM), and by the global workspace theory (GWT) of consciousness originated by cognitive neuroscientist Bernard Baars and further developed by him, Stanislas Dehaene, Jean-Pierre Changeux, George Mashour, and others. However, the CTM is not a standard Turing Machine. It’s not the input-output map that gives the CTM its feeling of consciousness, but what’s under the hood. Nor is the CTM a standard GW model. In addition to its architecture, what gives the CTM its feeling of consciousness is its predictive dynamics (cycles of prediction, feedback and learning), its internal multi-modal language Brainish, and certain special Long Term Memory (LTM) processors, including its Inner Speech and Model of the World processors. Phenomena generally associated with consciousness, such as blindsight, inattentional blindness, change blindness, dream creation, and free will, are considered. Explanations derived from the model draw confirmation from consistencies at a high level, well above the level of neurons, with the cognitive neuroscience literature. Reference. L. Blum and M. Blum, "A theory of consciousness from a theoretical computer science perspective: Insights from the Conscious Turing Machine," PNAS, vol. 119, no. 21, 24 May 2022. https://www.pnas.org/doi/epdf/10.1073/pnas.2115934119

n the ‘orthodox’ view, cognition has been seen as manipulation of symbolic, mental representations, separate from the body. This dualist Cartesian approach characterised much of twentieth-century thought and is still taken for granted by many people today. Language, too, has for a long time been treated across scientific domains as a system operating largely independently from perception, action, and the body (articulatory-perceptual organs notwithstanding). This could lead one into believing that to emulate linguistic behaviour, it would suffice to develop ‘software’ operating on abstract representations that would work on any computational machine. Yet the brain is not the sole problem-solving resource we have at our disposal. The disembodied picture is inaccurate for numerous reasons, which will be presented addressing the issue of the indissoluble link between cognition, language, body, and environment in understanding and learning. The talk will conclude with implications and suggestions for pedagogy, relevant for disciplines as diverse as instruction in language, mathematics, and sports.

Natural language is strongly context-dependent and can be perceived through different sensory modalities. For example, humans can easily comprehend the meaning of complex narratives presented through auditory speech, written text, or visual images. To understand how complex language-related information is represented in the human brain there is a necessity to map the different linguistic and non-linguistic information perceived under different modalities across the cerebral cortex. To map this information to the brain, I suggest following a naturalistic approach and observing the human brain performing tasks in its naturalistic setting, designing quantitative models that transform real-world stimuli into specific hypothesis-related features, and building predictive models that can relate these features to brain responses. In my talk, I will present models of brain responses collected using functional magnetic resonance imaging while human participants listened to or read natural narrative stories. Using natural text and vector representations derived from natural language processing tools I will present how we can study language processing in the human brain across modalities, in different levels of temporal granularity, and across different languages.

Language production is a form of behavior and as such involves executive control and the prefrontal function. The cognitive architecture of prefrontal executive function thus certainly plays an important role in shaping language production. In this talk, I will review the main features of the prefrontal executive function we have uncovered during the last two decades and I will discuss how these features may help understanding language production.

Executive functions refer to a collection of mental processes such as attention, planning and problem solving, supported by a frontoparietal distributed brain network. These functions are essential for everyday life. Specifically in the context of patients with brain tumours there is a need to preserve them in order to enable good quality of life for patients. During surgeries for the removal of a brain tumour, the aim is to remove as much as possible of the tumour and at the same time prevent damage to the areas around it to preserve function and enable good quality of life for patients. In many cases, functional mapping is conducted during an awake surgery in order to identify areas critical for certain functions and avoid their surgical resection. While mapping is routinely done for functions such as movement and language, mapping executive functions is more challenging. Despite growing recognition in the importance of these functions for patient well-being in recent years, only a handful of studies addressed their intraoperative mapping. In the talk, I will present our new approach for mapping executive function areas using electrocorticography during awake brain surgery. These results will be complemented by neuroimaging data from healthy volunteers, directed at reliably localizing executive function regions in individuals using fMRI. I will also discuss more broadly challenges ofß using neuroimaging for neurosurgical applications. We aim to advance cross-modality neuroimaging of cognitive function which is pivotal to patient-tailored surgical interventions, and will ultimately lead to improved clinical outcomes.

Abstract: There is no doubt that the study of brain damaged individuals has contributed greatly to our understanding of the mind/brain. Within this broad approach, cognitive neuropsychology accentuates the cognitive dimension: it investigates the structure and organization of perceptual, motor, cognitive, and language systems – prerequisites for understanding the functional organization of the brain – through the analysis of their dysfunction following brain damage. Significant insights have come specifically from this paradigm. But progress has been slow and enthusiasm for this approach has waned somewhat in recent years, and the use of existing findings to constrain new theories has also waned. What explains the current diminished status of cognitive neuropsychology? One reason may be failure to calibrate expectations about the effective contribution of different subfields of the study of the mind/brain as these are determined by their natural peculiarities – such factors as the types of available observations and their complexity, opportunity of access to such observations, the possibility of controlled experimentation, and the like. Here, I also explore the merits and limitations of cognitive neuropsychology, with particular focus on the role of intellectual, pragmatic, and societal factors that determine scientific practice within the broader domains of cognitive science/neuroscience. I conclude on an optimistic note about the continuing unique importance of cognitive neuropsychology: although limited to the study of experiments of nature, it offers a privileged window into significant aspects of the mind/brain that are not easily accessible through other approaches. Biography: Alfonso Caramazza's research has focussed extensively on how words and their meanings are represented in the brain. His early pioneering studies helped to reformulate our thinking about Broca's aphasia (not limited to production) and formalised the logic of patient-based neuropsychology. More recently he has been instrumental in reconsidering popular claims about embodied cognition.

Understanding how our brains process language is one of the fundamental issues in cognitive science. In order to reach such understanding, it is critical to cover the full spectrum of manners in which humans acquire and experience language. However, due to a myriad of socioeconomic factors, research has disproportionately focused on monolingual English speakers. In this talk, I present a series of studies that systematically target fundamental questions about bilingual language use across a range of conversational contexts, both in production and comprehension. The results lay the groundwork to propose a more inclusive theory of the neurobiology of language, with an architecture that assumes a common selection principle at each linguistic level and can account for attested features of both bilingual and monolingual speech in, but crucially also out of, experimental settings.

The world around us is complex, but at the same time full of meaningful regularities. We can detect, learn and exploit these regularities automatically in an unsupervised manner i.e. without any direct instruction or explicit reward. For example, we effortlessly estimate the average tallness of people in a room, or the boundaries between words in a language. These regularities and prior knowledge, once learned, can affect the way we acquire and interpret new information to build and update our internal model of the world for future decision-making processes. Despite the ubiquity of passively learning from the structured information in the environment, the mechanisms that support learning from real-world experience are largely unknown. By combing sophisticated cognitive tasks in human and rats, neuronal measurements and perturbations in rat and network modelling, we aim to build a multi-level description of how sensory history is utilised in inferring regularities in temporally extended tasks. In this talk, I will specifically focus on a comparative rat and human model, in combination with neural network models to study how past sensory experiences are utilized to impact working memory and decision making behaviours.

To better understand human scene understanding, we extracted features from images using CLIP, a neural network model of visual concept trained with supervision from natural language. We then constructed voxelwise encoding models to explain whole brain responses arising from viewing natural images from the Natural Scenes Dataset (NSD) - a large-scale fMRI dataset collected at 7T. Our results reveal that CLIP, as compared to convolution based image classification models such as ResNet or AlexNet, as well as language models such as BERT, gives rise to representations that enable better prediction performance - up to a 0.86 correlation with test data and an r-square of 0.75 - in higher-level visual cortex in humans. Moreover, CLIP representations explain distinctly unique variance in these higher-level visual areas as compared to models trained with only images or text. Control experiments show that the improvement in prediction observed with CLIP is not due to architectural differences (transformer vs. convolution) or to the encoding of image captions per se (vs. single object labels). Together our results indicate that CLIP and, more generally, multimodal models trained jointly on images and text, may serve as better candidate models of representation in human higher-level visual cortex. The bridge between language and vision provided by jointly trained models such as CLIP also opens up new and more semantically-rich ways of interpreting the visual brain.

It is generally agreed upon that qualities of conscious experience instantiate structural properties, usually called relations. They furnish a representation of qualities (or qualia, in fact) in terms of a mathematical space Q (rather than a set), which is crucial to both modelling and measuring of conscious experience." "What is usually disregarded is that “only such structural properties generalize across individuals” (Austen Clark), but that qualities themselves as differentiated by stimulus specifications, behavior or reports do not. We show that this implies that only the part of Q which is invariant with respect to the automorphism group has a well-defined referent, while individual elements do not. This poses a prima facie limitation of any theory or experiment that aims to address individual qualities. We show how mathematical theories of consciousness can overcome this limitation via symmetry groups and group actions, making accessible to science what is properly called private language.

A rich data representation that finds wide application in industry and research is the so-called knowledge graph - a graph-based structure where entities are depicted as nodes and relations between them as edges. Complex systems like molecules, social networks and industrial factory systems can be described using the common language of knowledge graphs, allowing the usage of graph embedding algorithms to make context-aware predictions in these information-packed environments.