Past events

Our ability to selectively attend to a particular conversation amidst competing input streams (e.g. other speakers) epitomized by the ‘Cocktail Party’ problem, is remarkable. How this demanding perceptual feat is achieved from a neural systems perspective remains unclear and controversial. In this talk I will present data from both invasive and non-invasive electrophysiological recordings in humans, investigating the manner in which selective attention governs the brain’s representation of attended and ignored speech streams using a simulated ‘Cocktail Party’ Paradigm. Results indicate that brain activity dynamically tracks speech streams using both low frequency phase and high frequency amplitude fluctuations, and that optimal encoding likely combines the two. In and near low level auditory cortices, attention ‘modulates’ the representation by enhancing cortical tracking of attended speech streams, but ignored speech remains represented. In higher order regions, the representation appears to become more ‘selective’. Furthermore, when to-be-ignored input has a predictable rhythmic structure, there is even evidence for active suppression of responses to these stimuli, making attention more effective. Viewing the facial movements of the speaker movements of a speech further enhances the selectivity of the neural response. Together, these findings are a testament to the proactive and flexible nature of the neural system which dynamically shapes its internal activity according to environmental and contextual demands.

Recent experiments indicate that primary auditory and visual cortex in rodents exhibit '"Salt and Pepper" architecture, consisting of highly selective neurons without columnar structure. Likewise, there is no apparent functional structure in the pattern of projections from the olfactory bulb to piriform cortex. In my talk I will address the questions: Can sharp stimulus selectivity be maintained in a cortical circuit with random connections? What are the computational ramifications of random cortical projections? How moderate tuning of cortical connectivity can be incorporated on top of largely random architecture? I will describe recent theoretical work that addresses these questions and will discuss their applications to sensory processing in rodent visual and olfactory cortices. I will also discuss relation between these results and recent developments in Machine Learning.

Numerous clinical and preclinical investigators have reported that in several important medical indications such as in (i) ischemic stroke, (ii) traumatic brain injuries (TBI), (iii) acute migraine cases, (iv) glioblastoma brain tumors and (v) epileptic attacks, there is a rapid accumulation in the brain of excess glutamate molecules which are excitotoxic and this leads to significant neurological damage and motoric incapacitations in patients. Vivian Teichberg introduced a method for scavenging of excess brain glutamate which consists of the intravenous administration of a recombinant preparation of the enzyme, Glutamate Oxaloacetate Transaminase (GOT). This causes a rapid decrease in blood glutamate levels and creates a gradient which leads to the efflux of the excess brain glutamate into the blood stream and reduces neurological damage. The main advantage of the Brain Glutamate Scavenging technology, over other drug treatments that are currently being developed, is that the augmentation of GOT activity occurs in the blood circulation and therefore, doesn’t affect normal brain neurophysiology, whereas the pharmacological inhibition of the activities of glutamate receptors or transport systems occurs in the brain, and could be followed by serious side effects in the central nervous system.

The spiking of dopamine neurons in animals, and apparently analogous BOLD signals at dopaminergic targets in humans, appear to report predictions of future reward. Prominent computational theories of these responses suggest that they both support and reflect trial-and-error learning about which actions have been successful, based on simple associations with past rewards. This is essentially a neural implementation of Thorndike's (1911) behaviorist principle that reinforced behaviors should be repeated. However, it has long been known that organisms are not condemned merely to repeat previously successful actions, but instead that even rodents' decisions can under some circumstances reflect other sorts of knowledge about task structure and contingencies. The neural and computational bases for these additional effects, and their interaction with the putative reinforcement systems in the basal ganglia, are poorly understood. Such interactions are of considerable practical importance because, for instance, disorders of compulsion in humans, such as substance abuse, are thought to arise from runaway reinforcement processes unfettered by more deliberative influences. I first discuss how such extra-reinforcement effects – e.g., planning novel routes based on cognitive maps, or incorporating "counterfactual" feedback about foregone actions – can be incorporated in the framework of existing computational theories, via algorithms for “model-based reinforcement learning." Rather than learning about actions' past successes directly, such algorithms learn a representation of the task structure, and can use it to evaluate candidate actions via mental simulation of their consequences. This computational characterization allows reasoning about (and explaining empirical data concerning) under which circumstances the brain might efficiently adopt either this strategy or the reinforcement one. It also allows quantifying and dissociating either strategy's effects on decision making and associated neural signaling. Next, I discuss human fMRI experiments characterizing these influences in learning tasks. By fitting computational models to decision behavior and BOLD signals, we demonstrate that neither choices nor (putatively dopamine-related) BOLD signals in striatum can be explained by past reinforcement alone, but instead that both reflect additional learning and reasoning about task structure and contingencies. That such influences are prominent even at the level of striatum challenges current models of the computations there and suggest that the system is a common target for many different sorts of learning. Additional experiments examine individual variation in the tendency to employ either system; the patterns of both spontaneous and experimentally induced variation suggest that the dominance of model-based decision influence over simpler reinforcement systems employs cognitive control mechanisms that have previously been studied in other areas of cognitive neuroscience. Finally, I report results showing that patients with several disorders involving compulsion show abnormally reinforcement-bound choices on our tasks, supporting a link between these neurocomputational learning mechanisms and pathological habits.