Abstract: Binding is an important process in human episodic memory. On retrieval, dual process theory posits familiarity and recollection as two separated processes that contribute to recognition memory. It has been hypothesized that item retrieval bases on familiarity and recollection, whereas relational retrieval employs recollection. ERPs studies on recognition memory have identified two old/new effects that respectively correspond to familiarity and recollection. The mid-frontal old/new effect, occurring at approximately 300~500ms over frontal electrode sites, is attributed to familiarity. The parietal old/new effect, approximately 500~800ms maximal over parietal electrode sites, reflects recollection. According to Van Pettern et al. (2002), there was also an executive process which was reflected by an old/new effect over prefrontal scalp around 700ms after stimulus onset.
In the present study, we explore the differences between relational retrieval and item retrieval by using the ERPs techniques and following the associative recognition paradigm. 16 college students (7 males, 9 females) took part in our experiment. They studied a series of face-verb pairs presented in sequence. In the test, some face-verb pairs were presented that were encoded together at study (intact), some were presented in which both items were studied but not together (rearranged), and the rest consisted of two new, unstudied items (new). Participants were instructed to determine whether a pair was “intact”, “rearranged”, or “new”. The ERPs for the three categories of correctly-judged responses in the test phrase were averaged separately. Thus we made our hypothesis that the differences between the ERPs of the “intact” and the “rearranged” pairs simply reflected relational retrieval. Also, in the early stage of item retrieval, item familiarity and relational recollection contributed to successful retrieval of the “intact”, whereas old/new effect only reflected the item familiarity of the “rearranged”.
It was found that old/new effects indicating the “intact” and the “rearranged” were at anterior cortex around 200ms after the stimulus, more obvious at all cortex from 300ms to 500ms, but prominent at prefrontal and frontal cortex from 700ms to 1400ms. However, relational old/new effect distributed over frontocentral-central- parietal areas in each time windows.
It could be inferred that item old/new effect and relational old/new effect occurred at the same time, the relational retrieval employing recollection occurred much later than item retrieval, and the late anterior item old/new effect could be related to executive processes of prefrontal cortex.

Key words: associative memory, item retrieval, relational retrieval, item old/new effect, relational old/new effect