Past events

Entropy and information are quantities of interest to neuroscientists, because of their mathematical properties and because they place limits on the performance of a neural system. However, estimating these quantities from neural spike trains is much more challenging than estimating other statistics, such as mean and variance. The central difficulty in estimating information is tightly linked to the properties of information that make it a desirable quantity to estimate. To surmount this fundamental difficulty, most approaches to estimation of information rely (perhaps implicitly) on a model for how spike trains are related. But the nature of these model assumptions vary widely. As a result, information estimates are useful not only in situations in which several approaches provide mutually consistent results, but also in situations in which they differ. These ideas are illustrated with examples from the visual and gustatory systems.

The octopus is an active hunter, a remarkable invertebrate whose complex behaviors rely on exceptionally good visual and tactile senses coupled with highly advanced learning and memory (LM) abilities. Studying the octopus LM system may therefore reveal characteristics universally important for mediation of complex behaviors. We developed slice and isolated brain preparations of the LM area in the octopus brain to characterize the short- and long-term neural plasticity. The importance of these processes for LM are been tested in behavioral experiments. The results support the importance of LTP in behavioral LM and suggest new ideas regarding the organization of short- and long-term memory systems.

I am interested in how neural dynamics, changes in neural sensitivity on the time scale of milliseconds to seconds, support rapid changes in perceptual capability. Our key focus is in testing the hypothesis that dynamics in early sensory cortices, such as the primary somatosensory cortex, play a key role in perception. To examine these issues in a model where detailed invasive studies are possible, we study the vibrissa sensory system of rodents. To examine the broader relevance of these findings, we have a smaller parallel effort I human perception and imaging. A variety of our studies suggest that. In this seminar, I will describe work we have done to understand the 'natural scene' of vibrissa perception, the signals that are generated during active surface contact. I will then describe how these signals are encoded in 2 recently discovered cortical maps, a frequency map, shaped by the resonance properties of the vibrissae, and a direction map, encoding the angle of vibrissa motion. I will then discuss why having these cortical feature columns, with adjacent neurons sharing similar computational processing, may enhance information processing. Specifically, I will argue that this arrangement facilitates dynamic regulation of neural sensitivity by neuromodulators that have a more course spatial precision, on the order of 300-1000 microns. I will briefly mention the hemo-neural hypothesis, the proposal that changes in blood flow and volume may be one source of neural regulation on this spatial scale. I will then describe an example of the dynamic regulation of spatial activation across a cortical map, resulting from adaptation during robust thalamocortical input at 5-25 Hz. I will argue that this adaptation leads to a transition into a state that is enhanced for discrimination of alternative stimuli, but impoverished for detection of novel stimuli (decreased sensitivity). To test the hypothesis that these kinds of cortical dynamics in the primary somatosensory cortex regulate perception, I will describe human MEG experiments showing that changes in the amplitude of SI activation regulate detection probability, and showing that these changes in perception and cortical amplitude are predicted by ongoing rhythmic activity in human SI at 5-25 Hz (the 'mu' rhythm).

Large, strongly coupled neural networks tend to produce chaotic spontaneous activity. This might appear to make them unsuitable for generating reliable sensory responses or repeatable motor patterns. However, this is not the case. Inputs can induce a phase transition, leading to responses uncontaminated by chaotic "noise". Likewise, appropriately trained feedback units can control the chaos, resulting in a wide variety of repeatable output patterns. These issues will be discussed accompanied by examples and demonstrations.