ESI Systems Neuroscience Conference 2021

The Natural Brain: understanding neuronal computation in its natural habitat

The visual brain of humans and other primates is evolutionarily adapted for the perception of important stimuli. This specialization is particularly obvious in the social domain, with an abundance of brain regions nominally devoted to the analysis of faces, bodies and actions. The face patch system, for example, is a collection of cortical islands, readily identified using fMRI, in which neurons respond more vigorously to images of faces than to those of other stimulus categories. While responses of the face patch system have been studied in detail, relatively little work has been directed to understanding its operation during naturalistic modes of stimulation and behavior - namely, the visual conditions under which the primate brain evolved. I will describe an intermodal mapping approach in which we investigate neural responses in the face patch system during the viewing of naturalistic videos. In contrast to most visual electrophysiology studies, which ask how neural responses encode or represent stimuli, we instead analyzed single cell responses based upon their functional relationship to fMRI activity elicited elsewhere in the brain during video viewing. Using this method, we found that face- selective neurons from four nodes of the face patch system fell into approximately ten functional groups, or subnetworks, each with a distinctive brain-wide signature of coactivation. Importantly, each functional subnetwork spanned multiple face patches, and each face patch contained a mixture of neurons from nearly all subnetworks. These findings offer a new view of the organization of the macaque face network and pose challenges for commonly made assumptions about segregation of function in cognitive neuroscience. The results also highlight the importance of testing the brain under conditions to which it is evolutionarily adapted, as well as the power of intermodal mapping as a tool to harness the complexity of natural behavior.

Naturalistic experimental paradigms in neuroimaging arose from a pressure to test the validity of models we derive from highly controlled experiments in real-world contexts. In many cases, however, such efforts led to the realization that models developed under particular experimental manipulations failed to capture much variance outside the context of that manipulation. The critique of non-naturalistic experiments is not a recent development; it echoes a persistent and subversive thread in the history of modern psychology. The brain has evolved to guide behavior in a multidimensional world with many interacting variables. The assumption that artificially decoupling and manipulating these variables will lead to a good understanding of the brain may be untenable. Recent advances in artificial neural networks provide an alternative computational framework to model cognition in natural contexts. In contrast to the simplified and interpretable hypotheses we test in the lab, these models do not learn simple, human-interpretable rules or representations of the world. Instead, they use local computations to interpolate over task-relevant manifolds in high-dimensional parameter space. Counterintuitively, over-parameterized deep neural models are parsimonious and straightforward, as they provide a versatile, robust solution for learning a diverse set of functions in natural contexts. Naturalistic paradigms should not be deployed as an afterthought if we hope to build models of brain and behavior that extend beyond the laboratory into the real world.

Our understanding of the hippocampus has been framed by two landmark discoveries: the discovery by Scoville and Millner that hippocampal damage causes profound and persistent amnesia and the discovery by O’Keefe and Dostrovsky of hippocampal place cells in rodents. However, it has been unclear to what extent spatial representations are present in the primate brain and how to reconcile these representations with the known mnemonic function of this region. I will discuss a series of experiments that have examined neural activity in the hippocampus and adjacent entorhinal cortex in monkeys performing behavioral tasks including spatial memory tasks in a virtual environment. These data demonstrate that spatial representations can be identified in the primate hippocampus, and that behavioral task structure has a significant influence on hippocampal activity, with neurons responding to all salient events within the task. Together, these data are consistent with the idea that activity in the hippocampus tracks ongoing experience in support of memory formation.

Our ability to measure behavior non-invasively is rapidly becoming a reality with deep learning-based approaches. In my talk, I will discuss our work on markerless pose estimation tools (such as DeepLabCut). I will outline the challenges and progress, and give our perspective on how such tools are allowing us to study the sensorimotor system with ever greater detail.

Animal behavior aims to reach an understanding of the instantaneous decision-making of individual animals throughout their lifetime. However, we are still far from such insights. Nevertheless, animal behaviorists have started a transformation recently by being able to quantify behavioral decisionmaking of individuals in their natural context in high temporal and spatial resolution throughout their life using onboard high-definition tracking devices as well as highest-definition tracking of shorter behavioral sequences in laboratory settings. One approach to potentially predict behavioral decisions is to abstract from the myriad of individualistic behavioral settings and rather to derive general tendencies of individual behavioral decisions. Specifically, I suggest that individuals try to continuously optimize three overarching necessities: 1. Their individual autonomy; 2. Their social interactions; 3. Their individual safety. In each species, individuals exist that have different priorities of these three drivers of behavior. Individual behavioral decision-making in whatever context (e. g. reproduction, foraging, migration) can be mapped onto these necessities and may thus explain previously poorly understood complex decisions. This mapping of generalized necessities onto immediate behavioral decisions also allows us to understand strategic decision-making of animals over longer time periods. Although there is currently no direct link between this explanatory framework and neurobiological systems, I will discuss possible synergies.

The collective behavior of organisms creates environmental micro-niches that buffer them from environmental fluctuations - thus coupling organismal-physiology and environmental-physics. This talk will illustrate several examples of this coupling, using honey bee swarms as a model system, wherein individual bees integrate physical information from their local environment to attain a global non-equilibrium steady state. Specific scenarios include bees who detect local temperatures to ventilate their hive on hot days, bees who collaborate to stabilize their reproductive swarm clusters, and bees that rally to their queen via a chemical ‘game of telephone’. Using a combination of biological experiments, theory, and computation, we aim to understand how environmental cues (mechanical forces, temperature, chemical concentrations) are converted to behavioral outputs (walking, fanning, scenting) that allow the bees to achieve dynamic homeostasis.