Abstract:  Previous research has identified robust collaborative inhibition and error pruning in collaborative memory: The collaborative group has worse recall performance but higher error pruning compared with the nominal group. However, nearly all the prior studies focus on the item recall task but fail to explore such two phenomena in the context retrieval task. Moreover, it is still unclear how emotion influences these two phenomena. In addition, the level of processing is also of our interest. Therefore, in current study we aimed to explore collaborative inhibition and error pruning in two experiments with different levels of processing by using stimuli with different valences in two recall tasks. Two experiments were designed identically, except the level of processing. Each experiment included forty-eight effective participants: Sixteen in the individual group and thirty-two in the collaborative group. The nominal group was constructed by randomly choosing two participants from the individual group. Three emotional valence conditions (positive, negative, and neutral) were set and each contained 20 two-character Chinese words, half red and half green. During the study phase, participants were asked to report the emotional valence of each word (deep encoding task) in Experiment 1, and to identify the displayed color (shallow encoding task) in Experiment 2. During the test phase, participants were instructed to recall the previously studied words (item recall) and the color of each word (context retrieval) individually or collaboratively. Two tasks (item recall versus context retrieval) in Experiment 1 that using deep encoding task revealed robust collaborative inhibition and error pruning but in different patterns: Collaborative inhibition was stronger in item recall, while error pruning was higher in context retrieval. The interaction between recall task and emotional valence was significant in collaborative inhibition: Inhibition in emotional words was stronger than neutral words in item recall, but no difference was found in context retrieval. Results in Experiment 2 that using shallow encoding task also showed significant collaborative inhibition and error pruning, but these two phenomena did not vary significantly in the task type. To conclude, different patterns between the two recall tasks of collaborative inhibition and error pruning showed in deep level of processing give support to the dual-process model. The interaction between recall task and emotional valence in both experiments strengthens the view of trade-off, which also reinforces the dual- process model. The modulation of recall task on the two phenomena is also influenced by level of processing.

Key words:  collaborative inhibition, error pruning, item recall, context retrieval, emotional valence, level of processing