Understanding failures of learning: Hebbian learning, competition for representational space, and some preliminary experimental data.

Publisher Summary This chapter discusses the reason of difficulty in paired associate learning in amnesics. In an experiment described in the chapter, the subject receives a list of, say, 12 word pairs (including, for example, locomotive-dishtowel and table-banana, among others). After a slight delay, the experimenter presents the first word in one of the pairs, and asks the subject to recall the word that was previously paired with it in the experiment. Because of the subject's amnesia, the subject may not remember even that there was a list of word pairs. Nevertheless, as is standard in paired-associate learning, the subject is encouraged to guess a response. Given the arbitrary pairing of the words, table is unlikely to come to mind in the context of banana as a cue, and hence the stimulus is likely to elicit some other response. If learning is Hebbian, it is this response that will be strengthened, thereby leading to interference. There is experimental support for the idea that forcing amnesics to make their own responses to items leads to interference.