Prediction suppression in monkey inferotemporal cortex depends on the conditional probability between images.

Area of Science:

• Neuroscience
• Cognitive Science
• Animal Behavior

Background:

• Neurons in area TE of the inferotemporal cortex exhibit prediction suppression when monkeys view images in a fixed sequence.
• Prediction suppression is a phenomenon where a repeated trailing image elicits a weaker neural response.
• This suppression may depend on image contiguity (co-occurrence) or contingency (correlation).

Purpose of the Study:

• To differentiate whether prediction suppression relies on image contiguity or contingency.
• To investigate the role of joint probability (P(A,B)) versus conditional probabilities (P(A|B), P(B|A)) in prediction suppression.

Main Methods:

• Monkeys were trained on image sequences under regimens that kept joint probability constant but varied conditional probabilities.
• Prediction suppression was measured during subsequent testing after the training period.
• Statistical measures like joint probability P(A,B) and conditional probabilities P(A|B) and P(B|A) were used.

Main Results:

• Reducing either P(A|B) or P(B|A) during training significantly attenuated prediction suppression.
• The degree of prediction suppression correlated with the strength of the predictive relationship between images.

Conclusions:

• Prediction suppression is dependent on the contingency between images, reflecting learned predictive relationships.
• Contiguity alone (co-occurrence) is insufficient to induce robust prediction suppression.
• This suggests that the brain actively learns and relies on probabilistic predictive information in visual sequences.