Gustatory processing changes within sessions, dependent on attention

By Donald B. Katz 

Gustatory cortical networks reliably transition through taste-specific states

By Lauren M. Jones 

Distinct states of firing patterns in the primary visual cortex of awake ferrets

By József Fiser 

In a series of three talks, we will present a coherent view of brains functioning as a dynamic network processing complex multi-dimensional input. The specifics of this approach include a) large scale multi-electrode recordings as the appropriate tool for data collection, b) performance of behavioral tasks as the appropriate preparation, c) data analysis methods based on the principle that processing in such networks depends on the prior state of the networks and involves complex coherent patterns of activity in neural ensembles, and d) the use of multiple modalities and species for validation of the generality of results.  In the first talk, we show that the processing of tastes in gustatory cortex varies across a recording session dependent on mass action in cortical networks, such that as the rat switches from being attentive (desynchronized EEG) to being inattentive (oscillatory EEG) the nature of taste coding changes; in inattentive rats, neural and behavioral responses to a battery of experimenter-administered tastes are coded more simply in relation to palatability—pleasing tastes are more distinct from noxious tastes, and noxious tastes are more similar to each other. In the second talk, we show that taste responses vary from trial to trial, dependent on coherent ensemble processes. Ensembles of taste cortex neurons go through a series of coherent states—defined as a period of time in which each neuron has a particular firing rate—with minimal switching time between states; a particular series of 2-3 states is specific for a particular taste stimulus, but the time spent in each state varies widely across trials. A given state series is a more effective predictor of taste identity than PSTH-based methods. Finally, in the third talk we switch to a new modality (vision), a more unconstrained task (passive viewing), and a different species (ferret). While the visual system that is thought to work very differently from taste, and the passive viewing task does not require stimulus processing and response (ferrets sat in complete darkness or watched either a natural scene or a random noise movie), cortical functioning still exhibits similar hallmarks of dynamic state-dependent functioning. We suggest that rather than attempting to reproduce simple output or trial-averaged responses of single neurons, analyses that capture the underlying structure of the dynamic behavior may provide useful information for theoretical/computational attempts to unravel the function of cortical networks.