Abstract
To survive, organisms must extract information from the past that is relevant for their future. How this process is expressed at the neural level remains unclear. We address this problem by developing a novel approach from first principles. We show here how to generate low-complexity representations of the past that produce optimal predictions of future events. We then illustrate this framework by studying the coding of ‘oddball’ sequences in auditory cortex. We find that for many neurons in primary auditory cortex, trial-by-trial fluctuations of neuronal responses correlate with the theoretical prediction error calculated from the short-term past of the stimulation sequence, under constraints on the complexity of the representation of this past sequence. In some neurons, the effect of prediction error accounted for more than 50% of response variability. Reliable predictions often depended on a representation of the sequence of the last ten or more stimuli, although the representation kept only few details of that sequence.
Highlights
Organisms often operate in unknown and uncertain environments
We find that for many neurons in primary auditory cortex, trial-by-trial fluctuations of neuronal responses correlate with the theoretical prediction error calculated from the shortterm past of the stimulation sequence, under constraints on the complexity of the representation of this past sequence
By using a basic principle from information theory, we show how to derive explicitly the tradeoff between quality of prediction and complexity of the representation of past information. We apply these ideas to a concrete case–neuronal responses recorded in auditory cortex during the presentation of oddball sequences, consisting of two tones with varying probabilities
Summary
Organisms often operate in unknown and uncertain environments. Extracting aspects of past observations, which are maximally predictive of the relevant future, is essential for survival. It has been suggested that the sensory cortex evolved to extract the statistical regularities of the world [1]. Adaptation of the nervous system to the statistical structure of the input is reflected in studies of neuronal responses to natural stimuli. In the auditory system, auditory nerve fibers–part of the auditory periphery–have been shown to achieve high coding efficiency by implementing a “tuned” nonlinear filter that selectively amplifies the anticipated signal [2]. In the visual system, Laughlin [3] showed that the contrastresponse function of interneurons in the fly's compound eye approximates the cumulative probability distribution of contrast levels in natural scenes
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have