Occipital EEG was monitored while subjects inspected 27 projected patterns. The number (N) and variety (V) of elements in the patterns were varied systematically. There were three levels of N (6, 12 or 24 elements) and three levels of V (circles, squares or hexagons occupying all, one half or one third of the element locations for all levels of N). Subjects were instructed that they would be required in a post-test to recognize the patterns, among patterns which had not appeared; they were also informed that the patterns had been constructed according to a set of simple rules, but the nature of these rules was not made fully explicit. The EEG was quantified by means of low-frequency analysis, yielding measures of abundance (theta, alpha and beta) and mean dominant frequency. For the recognition task, nine stimulus items were embedded among 45 items. Recognition efficiency was measured by means of the signal detection theory discrimination index (d'). The results were as follows: (i) Both N and V were inversely related to alpha abundance (P < 0.01); (ii) the strongest relationship between stimulus parameters and the EEG held for N and EEG beta activity (13.5–19.5 Hz; P < 0.001), where again the EEG and N were inversely related; (iii) there was a significant (P < 0.05) direct relationship between N and theta activity; (iv) contrary to prediction, mean dominant alpha frequency decreased as 'N increased; (v) d' correlated significantly with a number of effects for N, i.e. subjects who exhibited greatest EEG discriminability of items during exposure of the patterns, subsequently obtained the higher detection scores in the recognition task. The work described therefore demonstrates that not only do stimulus parameters have systematic effects upon brain activity as measured by the EEG, but that such effects have functional value and reflect aspects of efficiency. The results are fully compatible with arousal theory constructs relating physiological reactivity and performance.