A Study Of Proactive Interference Effect In Working Memory

Working memory is central to successful cognition. As proactive interference (PI) is one of the critical factors which exert significant control over the amount of information that can be retrieved from working memory, it is important to investigate the effect of distractor representation and how can we overcome this negative effect.This study includes three experiments. Working memory selection task was used in experiment one and the competition between target stimuli representation and distractors representation was fully examined through a series of behavioral tasks. Participants were told to remember a set of letters in the study display and later were cued to select half of the letters and to forget the others. The selected letters were the target stimuli and were called as the to-be-remember (TBR) set and the other letters were distractors and were named as to-be-forgotten (TBF) set. Soon after the selection cue, the probe was displayed on the screen and participants had to make a"yes/no"judgment as to whether the probe was one of the three target stimuli.We modulated the types of selection cues and the cue-to-probe intervals to test selection difficulty in working memory. It was found that when selection cues represented target stimuli, the target representation became more active and made it easier for the target to be selected. However, even though the level of activation of the distractor representation increased with cues reinstating the TBF set, it did not make selection more difficult. These results provided partial support to the relative activation hypothesis-- the level of target representation had a significant impact on selection difficulty, that of the distractor representation did not.Using event-related potential (ERP) technique, experiment two and three investigated the eletrophysiological signature of PI in directed-forgetting task and recent probes task, respectively. Directed-forgetting task equals with TBF cue condition in experiment one. The probe was selected in different ways in this study. The probes were in the memory set for the"Yes"trials, in the recently ignored set for the"Lure"condition and out of the initial memory set for"No"trials. Proactive interference was indexed as delayed response to probes in the Lure condition than in the No condition and differences between Lure and No waveforms. In the recent probes task, participants were presented with a study display including four letters. They were to judge whether the following probe was in the study display or not. To manipulate the familiarity of the probes, each trial was composed of 2 letters from the previous trial and 2 letters that had not been presented in the previous two trials. This allowed some probes were members of the current trial and the last trial (Recent Positive probes); some were members of the current trial but not the last two trial (Non-Recent Positive probes); some were members of the previous trial but not the current trial (Familiar Negative probes); and some had not been presented in the current trial and the previous two trials (Non-familiar Negative probes). Proactive interference effect was thought to be indexed by differences between the two kinds of negative probes in this study.As the results of experiment two showed, error rate was increased and reaction time was delayed for the Lure condition relative to No condition. A fronto-central N2 component peaking around 300 ms post-probe-onset differentiated among different types of probes, with Lure N2 significantly reduced when compared with No condition. There was no difference between P3 amplitudes, but P3 latency was earlier for Yes condition than for Lure and No condition. The study identifies N2 as the ERP signature for proactive interference resolution. It also indicates that the resolution process occurs in the same time window as target/non-target discrimination and provides the first piece of electrophysiological evidence supporting biased competition model based on localization data.The anterior N2 and P3 component were consistently found in experiment three. N2 component differentiated between yes/no responses, with yes trials induced reduced N2. There was no difference between Familiar and Non-familiar positive trials and between Familiar and Non-familiar negative trials. Moreover, the latencies of P3 components were earlier for positive trials than for negative trials. Further comparison of P3 amplitudes between Familiar and Non-familiar negative conditions found decreased amplitude for Familiar negative trials. We inferred the modulation of P3 amplitudes indexed proactive interference resolution.Combining evidences from experiment two and experiment three, it is postulated that the anterior N2 and P3 component correspond to the two processes in the dual-process theory, i.e., judgment of familiarity and information collection. Dual-process theory assumes that recognition reflects a mixture of judgments based on familiarity and recollection of specific episodic information. The two tasks used in ERP studies have different memory load and different level of distractor representation activation. Maybe these differences lead to different timing course of online processing of PI. The ERP components reflecting PI resolution had reduced amplitudes in experiment two and experiment three, indicating the resolution of PI does not need inhibition with extra mental effort. The electrophysiological evidences are consistent with results from experiment one, that is, successful selection in working memory depends on increased activation of target representation but not decreased activation of distracor representation.