The present study compares brain responses to structured and unstructured tone sequences from the viewpoint of early Gestalt psychology.
In a first experiment, a tone sequence with elements randomly permuted in time was contrasted to such one in which a clearly recognizable rhythmical pattern was enclosed. In a second experiment, a 'melodic Gestalt' based on traditional musical rules was compared to a random version.
In the rhythm experiment, 40 structured and 40 unstructured examples were used. The structured versions included one out of four rhythmical patterns (quarter notes, syncopation, dotted rhythm or a group of an eighth note and two sixteenth notes). Each sequence consisted of 60 tones of constant pitch. In the melody experiment, 40 well-balanced melodies and 40 randomized sequences were presented, consisting of identical elements but in different order. In the randomized sequences, pause length between tones was the same (75 msec) and the mean pause length between conditions was kept constant.
In each experiment, 14 non-musicians listened attentively to the examples. In the rhythm experiment, grand average ERPs over all subjects, all sequences and individual tones revealed a sustained negative potential between 220 and 700 msec for the random sequences, but not for the rhythmic patterns. Amplitude maxima were at frontal and central electrodes. In the melody experiment, grand average ERPs showed significantly increased N1 and P2 components for the melodies and a P3- like component for the random sequences.
In both experiments, ERP components suggest an ongoing comparison of the actual input with the preceding sound information: In the rhythm experiment, the sustained negative potential might indicate stronger mental effort for the processing of irregular compared to regular sequences. As long as no clear rhythmical pattern is recognized, mental effort is needed to integrate a number of recent tones in order to detect possible rhythmic regularities. Once the pattern is found, task load is low and strong predictions on the following tones are confirmed each time. In the melody experiment, a higher prediction for tones in structured sequences seems to elicit significantly increased N1 and P2 components. Thus, auditory processing might be influenced by a top-down activation of melodic concepts already in this early stage of stimulus encoding.
In summary, similar paradigms yield different ERP results for melodic and rhythmic Gestalts. The fact might be explained by the different complexity of both patterns, or by a fundamentally different way of Gestalt processing in the two domains.