Judged: two independent readers; copy check clear.
on-center/off-surround
on-center/off-surround In this learning model, on-center/off-surround describes a recurrent network pattern in which the strongest or most predictive sequence representation is driven into activity while competing, less predictive representations are suppressed.
In this learning model, on-center/off-surround describes a recurrent network pattern in which the strongest or most predictive sequence representation is driven into activity while competing, less predictive representations are suppressed. This supports chunking by helping the system select the item groupings that best fit the sequence currently held in working memory.