- Differential effects of high memory load and high item similarity on short-term recognition : an event-related potential study (2011)
- Working memory (WM) contributes to countless activities during everyday live: reading, holding a conversation, making tea and so on. The core processes of WM comprise the phases of encoding, maintenance and retrieval. Successful recognition of stored objects requires several subprocesses such as stimulus encoding and evaluation, memory search and the organisation of a decision and a response. Much research has focused on encoding and maintenance of information but little interest has been directed to the retrieval of information. This is why the present dissertation investigated the neuronal correlates of retrieval of previously stored information and its modulation by load and probe-item similarity. Here memory load and probe-item similarity were manipulated in order to investigate the neuronal correlates of the recognition process using electroencephalography (EEG). We tested the hypothesis recognition is influenced differently by probe-item similarity and by memory load and that these factors are re Effected by distinct neuronal correlates. Furthermore we tested whether distinct neuronal responses could be related to a summed similarity model. The analysis of high-density ERP recordings showed both a load effect (load 1>load 3) and a similarity effect In addition, there was an interaction between load and similarity. The load effect was present during the whole epoch and did not change over time, whereas the similarity effect showed two distinct components between 300-600ms. In contrast to the load effect the similarity effect changed its sign over time. For the rest component, match probes elicited the strongest ERP responses, whereas for the second component dissimilar probes yielded the strongest ERP responses. The timing of the similarity effect corresponded well with the early and late P3b complex. The P3b complex is associated with stimulus categorisation and evaluation (early subcomponent) and memory search and criterion testing (late subcomponent). The results suggest that the difficulty of a task is not only determined by load but also enhanced by probe-item similarity. Since increasing the number of samples (i.e. memory load) can also increase the probe-item similarity (i.e. the probability that one of the samples is perceptually similar to the probe), an independent manipulation of both factors is indispensable to disentangle their particular impact on short-term recognition. Furthermore, I propose that the two distinct neural correlates of the P3b complex reeffects different stages of task processing connected with probe-item similarity. As suggested by summed similarity VI models, these components might re ect the subprocesses of similarity summation (early P3b) and criterion testing (late P3b).