Effect Of Example Variability On The Implicit Learning Of Multiple Nonadjacent Rule

A wealth of studies has confirmed the importance of example variability in explicit learning.However,the role of example variability in implicit learning,another powerful human learning mechanism,remained a largely open question.The present study explored this issue by manipulating the example variability of the non-target and target features using the multiple nonadjacent rules that were defined over the tones with which Chinese syllables were spoken.Studying the effect of the example variability on implicit learning will offer a new perspective to investigate the nature of example variability learning and provide a new direction to optimize implicit learning.Taken together,three experiments were conducted.Experiment 1 explored how implicit learning was affected by the example variability of the non-target feature(i.e.,syllables)by manipulating three conditions(zero variation,small variation,and high variation).The results showed that the participants implicitly learned the multiple nonadjacent rules in the high variation and zero variation conditions but not in the small variation condition.That is,the implicit learning performance across the three variability conditions showed a U-shaped curve.However,in Experiment 1,the training and test phase materials included all the mapping relations between the ping tones(tones 1 and 2)and ze tones(tones 3 and 4).It failed to clarify whether the knowledge was implicitly acquired was the underlying abstract rules(i.e.,ping was mapped to ze)or the specific mapping of some tones(e.g.,the mapping relation of tone 1 and tone 3).As such,the relation between non-target feature variation and implicit learning does not appear to be clear.Therefore,Experiment 2 set the transfer task by manipulating the degree of abstraction of the mapping rule to further investigate whether people can implicitly learn the rule per se and the role of example variability in this process.Specifically,in Experiment 2a,participants were asked to learn a set of strings that only contained specific mapping of some tones and then required to complete two classification tasks.One task used the same specific mapping as the training phase(non-transfer test),and another task had a greater degree of abstraction which used all the mapping relations between the ping and ze tones(transfer test).In Experiment 2b,participants were asked to complete another transfer task using a specific mapping completely different from the training phase.The results showed that there was a transfer effect of implicit learning in the zero variation and high variation conditions.Conversely,in the small variation,participants performed better than chance only in the non-transfer test.These findings indicated that the underlying abstract rules could be implicitly learned and which were influenced by example variability.Experiment 3 further explored the effect of example variability of the target feature(i.e.,tone type strings)on implicit learning by manipulating three variation conditions(zero,small and high).We found that although there was a diversiform non-target feature of the example,only one unvaried target feature was insufficient for participants to extract the rule implicitly(i.e.,zero variation condition).With the example variability of the target feature increase,participants could implicitly learn the rule(i.e.,small and high variation conditions).These findings suggested that the example variability of the target feature may play a more critical role in the implicit learning of multiple nonadjacent rule than the varied non-target feature of the example.The above results provide new evidence regarding the contribution of example variability to implicit learning.In conclusion,our study indicated that(1)the implicit learning performance across the example variability of non-target feature showed a U-shaped curve.These features with unvaried or highly varied lead to better implicit performance.(2)The implicit acquisition of the underlying rules rather than the specific mapping of some tones was affected by the example variability of the non-target feature.In parallel,these results provided direct evidence for the abstract question of implicit learning.(3)Compared to the unvaried target feature of the example,a certain number of variables along these features could implicitly extract the rule successfully.Summarily,these findings could provide a suggestion to obtain the optimal variability conditions for high-efficiency implicit learning.