THE TIME-COURSE OF RECOGNITION OF NOVEL MELODIES

Seven experiments explored the time course of recognition of brief nov el melodies. In a continuous-running-memory task, subjects recognized melodic transpositions following delays up to 2.0 min. The delays were either empty or filled with other melodies. Test items included exact transpositions CT), same-contour lures (SC) with altered pitch interv als, and different-contour lures (DC). DCs differed from Ts in the pat tern of ups and downs of pitch. With this design, we assessed subjects ' discrimination of detailed changes in pitch intervals (T/SC discrimi nation) as well as their discrimination of contour changes CT/DC). We used both artificial and ''real'' melodies. Artificial melodies differ ed in conformity to a musical key, being tonal or atonal. After empty delays, T/DC discrimination was superior to T/SC discrimination. Surpr isingly, after filled delays, T/SC discrimination was superior to T/DC . When only filled delays were tested, T/SC discrimination did not dec line over the longest delays. T/DC performance declined more than did T/SC performance across both empty and filled delays. Tonality was an important factor only for T/SC discrimination after filled delays. T/D C performance was better with rhythmically intact folk melodies than w ith artificial isochronous melodies. Although T/SC performance improve d over filled delays, it did not overtake T/DC performance. These resu lts suggest that (1) contour and pitch-interval information make diffe rent contributions to recognition, with contour dominating performance after brief empty delays and pitch intervals dominating after longer filled delays; (2) a coherent tonality facilitates the encoding of pit ch-interval patterns of melodies; and (3) the rich melodic-rhythmic co ntours of real melodies facilitate T/DC discrimination. These results are discussed in terms of automatic and controlled processing of melod ic information.