Fig 1.
To test how music expectancy modulates parallel visual learning, we implemented a factorial design across two experimental variables–music familiarity and music regularity. Each variable had two conditions, making four music conditions in total. A: learned regular music; B: learned irregular music; C: unlearned regular music, D: unlearned irregular music.
Fig 2.
Music learning and music condition validation.
(A) Final music memory retrieval performance (36 music stimuli in total): The task returned a retrieval score from 0–8. The plot shows the accuracy distribution for each music condition (excluding the control condition, which cannot have errors/be tested). (B) (C) key clarity and pulse clarity were measured using MIRtoolbox from MATLAB for both regular and irregular stimuli. Pair-wise comparisons (t-test) were conducted to compare regular and irregular music in these two features. Each pair of regular and irregular music is connected using lines. T-tests showed significantly higher pulse clarity in regular music but only a trend difference in clarity.
Fig 3.
(A) Two-day task procedure (B) Visual encoding paradigm used on Day 2: this is the main task phase during which participants repeatedly learned visual sequences paired with music. (C) notes and waveforms for example regular music stimuli and the irregular version of it. (D) example visual shape stimuli.
Fig 4.
This figure represented the cumulative learning curve during visual encoding. The plot illustrates the average proportion of visual sequences learned so far at each phase/run as a function of each music condition. The significantly slowest learning happened in the learned-irregular condition. To simplify the visualization, we combined learned control with unlearned control condition performance for each phase because 1) there were no statistically significant differences between them (Tukey’s HSD: p = .9113) plus 2) conceptually, the monotonic sequences, without dynamic changes in notes and temporal intervals, could not be “learned” and schematized thus were identical in both familiar and unfamiliar conditions.
Fig 5.
Visual sequence retrieval performance.
The bar plot represents average retrieval accuracy for visual sequences comparing learned vs. unlearned, three regularity conditions. Average retrieval accuracy for each condition from left to right is: 0.901, 0.93, 0.917, 0.888, 0.84, 0.926. Standard deviations from left to right are: 0.299, 0.255, 0.277, 0.316, 0.357, 0.27. Tukey’s HSD pair-wise comparisons indicated that the most difference came from the learned-irregular group.
Fig 6.
Visual retrieval task–response time for correctly trials only.
The bar plot represents average response times for correctly retrieved trials only for visual sequences comparing learned vs. unlearned, three regularity conditions. Average reaction time for each condition from left to right is: 5.74, 5.33, 5.9, 5.8, 5.57, 5.41. Standard deviations from left to right are: 2.05, 1.83, 2.06, 2.14, 1.95, 1.9.