Brain potentials, perceptual mechanisms and semantic categorisation

The behaviour of intermediate and long-latency evoked potential (EP) components were investigated in a semantic categorisation task. Visually presented 'narrow' and 'wide' priming category questions, intended to vary processing load, were followed, after 1.5 sec by a 100 msec presentation of a positive or negative instance for decision. Neither behavioural response latencies nor amplitudes of N1-P2 (130 and 212 msec mean latencies) differentiated between 'narrow' and 'wide' category decisions. However, response latencies were faster and N1-P2 amplitudes greater for positive than for negative instances of primed categories. This result is attributed to the priming effect, by 'spreading activation, of prior category questions on 'logogen-like' pre-conscious word detecting devices within the brain. Semantic discrimination by N1-P2 contradicts previous assertions that intermediate components only reflect selection based on crude stimulus characteristics. Comparison of N1-P2 amplitudes at lateral electrodes offered no evidence of hemispheric lateralisation of language function in the experimental task. There was no late wave unequivocally identifiable as P300 on morphological and scalp distribution criteria. A positive peak at mean latency 327 msec (called P3) was of low amplitude and stable latency and did not appear to index processing load as defined by the experimenters. However, the amplitude of N2-P3 was larger to positive instances of narrow categories than any of the other three conditions. This could be the sort of probability effect seen in P300 behaviour if it is assumed that the brain can discriminate between stimuli on the basis of the size of the subset from which they come.