The Language of Neurons15th & 16th September 2023, Barcelona

The brain must efficiently orchestrate distributed computation across the whole brain in order not only to survive but also to thrive. This implies a hierarchical brain orchestration of state-dependent, self-organised brain activity over time facilitating the necessary near optimal information transfer and processing at the lowest metabolic cost. Here we describe a theory of brain function providing a natural, parsimonious way to explain the underlying computational mechanisms for hierarchical brain orchestration. This theory was inspired by thermodynamics which has been a highly successful framework for characterising hierarchical transformations of physical systems including those found in systems biology. ‘Thermodynamics of Mind’ theory carefully combines model-free precise space–time hierarchy measures with causal generative whole-brain modelling to provide solid causal inference for the underlying brain mechanisms. Overall, implementations of the ideas of this novel theory have already provided insights into the principles underlying changes in orchestration and hierarchical brain organisation. This holds great promise for revealing the breakdown of hierarchy in neuropsychiatric disease and novel ways to rebalance.

Everyday activities shape our brains’ connectivity. We extract meaning from and assign meaning to both objects and symbols. A particularly interesting example are number symbols that refer to arbitrary sets of objects. These non-symbolic small and large quantities activate the two brain hemispheres differently, eventually leading to systematic spatial associations for number symbols. I will selectively review the evidence for this association and its relevance for mental arithmetic.

The proposed mechanism for spatial associations as inherent part of symbol meaning informs our view on the grounding, embodiment and situatedness of cognition.

In the last decades technological advancements providing new tools to study the brain in humans and non-human animals boosted the enormous progress of Neuroscience. This progress brought along a diversity of approaches, levels of description and the focusing on different granularities, so that a unified theory of brain function and its relationship to behavior is currently not available. Thus, the notion of neural representation is a topic of ongoing debate and controversy within the field of cognitive neuroscience. The relevant literature includes very diverse theoretical approaches, ranging from the neo-phrenological approach, assigning to specific brain modules specific cognitive functions, to dynamic system theory-inspired approaches, holding that brain neural activity emerges from the interactions among interconnected neurons, rather than being solely determined by fixed neural representations.

In the present talk the notion of neural representation will be addressed from an embodied perspective, discussing embodied simulation theory within the framework of neural reuse. It will be argued that being, feeling, acting, and knowing describe different modalities of our relations to the world, all sharing a constitutive underpinning bodily root that maps into dynamic ways of functioning of the brain-body. Brain architecture does not respect the boundaries of standard mental terms and categories. Mental terms and categories are the loose and imprecise verbal descriptions of a variety of complex cognitive behaviors that we are not yet able to fully explain in terms of their underlying neural and bodily mechanisms.

Understanding how the brain uses information is a fundamental goal of neuroscience. Several human disorders (ranging from autism spectrum disorder to PTSD to Alzheimer’s disease) may stem from disrupted information processing. Therefore, this basic knowledge is not only critical for understanding normal brain function, but also vital for the development of new treatment strategies for these disorders.

Memory may be defined as the retention over time of internal representations gained through experience, and the capacity to reconstruct these representations at later times. Long-lasting physical brain changes (‘engrams’) are thought to encode these internal representations. The concept of a physical memory trace likely originated in ancient Greece, although it wasn’t until 1904 that Richard Semon first coined the term ‘engram’. Despite its long history, finding a specific engram has been challenging, likely because an engram is encoded at multiple levels (epigenetic, synaptic, cell assembly). My lab is interested in understanding how specific neurons are recruited or allocated to an engram, and how neuronal membership in an engram may change over time or with new experience. Here I will describe data in our efforts to understand memories in mice.

This talk explores the concept of neural representation through the lens of active inference, a normative framework for understanding brain function. It delves into how living organisms employ generative models to minimize the discrepancy between predictions and observations (or variational free energy), emphasizing the role of generative models and belief dynamics. The talk highlights that the brain learns generative models to navigate the world adaptively, not (or not solely) to understand it. Different living organisms may possess an array of generative models, spanning from those that support action-perception cycles to those that also enable planning and imagination; from “explicit” models that involve variables for inferring external entities like objects, faces, or people, to “action-oriented models” that prioritize predicting action outcomes, omitting the need to encode external entities. Then, it elucidates how generative models and belief dynamics might link to neural representation and the implications of different types of generative models on understanding an organism’s ecological niche and cognitive capabilities. The talk concludes with open questions regarding the evolution of generative models and the development of advanced cognitive abilities – and the gradual transition from “pragmatic” to “detached” neural representations.

How do we represent objects and concepts in our minds and brains? This question has been considered nonsense. It has also been claimed that representations, in a narrow sense, do not exist, but that dynamic activity patterns explain what we envisage as the ‘mental images’ of the world. But others have postulated that unique representational ‘nodes’ exist for each object, concept or referring symbol, and even for their perceptual and conceptual features.

How can we find out? I propose to try it out. By creating networks of neuron-like elements, by connecting these as they are in local cortical circuits, areas, and, at a larger scale, in the human connectome, by implementing regulation and control mechanisms and, most importantly, by realistic learning procedures. Such “brain-constrained” neural networks can be treated like infants who experience objects with varying degree of similarity and later-on words and larger chunks of language in their context. We can then ask what goes on in the brain-like architectures when they experience the world and when they learn symbols.

One result of this endeavor is (of course) that the answer very much depends on the brain-constraints implemented. Networks with sequential area links, but no within-area connections, build fully distributed dynamic patterns. Adding reciprocal between-area links, within-area excitatory connections and local inhibition leads to circuit formation. A circuit includes many neurons and may be distributed across different network areas. Importantly, it may ‘ignite’ as a whole and maintain its reverberant activity for some time.

The talk will report on some recent results from simulating – and possibly explaining at a neurobiological level – spontaneous concept formation, as it is seen in preverbal infants. Symbol learning will be addressed, as it normally occurs from the end of the first year of life, along with the fast mapping of symbol form to meaning. Learning-simulation experiments will address different symbol types along with questions about differences between different types of referring expressions and between concrete and abstract concepts and symbols.