Associative recognition: A case of recall-to-reject processing

Abstract

Two-process accounts of recognition memory assume that memory judgments are based on both a rapidly available familiarity-based process and a slower, more accurate, recall-based mechanism. Past experiments on the time course of item recognition have not supported the recall-to-reject account of the second process, in which the retrieval of an old item is used to reject a similar foil (Rotello & Heit, 1999).In three new experiments, using analyses similar to those of Rotello and Heit, we found robust evidence for recall-to-reject processing in associative recognition, for word pairs, and for list-discrirnination judgments. Put together, these results have implications for two-process accounts of recognition.One of the key issues in recognition memory research is the nature of underlying processing, such as whether there is more than one process involved and what each process accomplishes. According to single-process accounts of recognition (e.g., Estes, 1994;Gillund & Shiffrin, 1984;Heit, 1993;Hintzman, 1988;Nosofsky, 1988), judgments are based on a unitary assessment of strength, corresponding to the summed similarity of the test item to all the relevant items in memory. Single-process memory models have been shown to account for a wide variety of data (see Clark & Gronlund, 1996, and Raaijmakers & Shiffrin, 1992, for reviews).A growing body of research, however, suggests that recognition judgments may involve a second, recall-like process (e.g., Atkinson & Juola, 1973Dosher, 1984;Gronlund & Ratcliff, 1989;Hintzman & Curran, 1994;Humphreys, 1978;Jacoby & Dallas, 1981;Jones & Heit, 1993;Mandler, 1980;Ratcliff & McKoon, 1982, 1989Yonelinas, 1994; see Mandler, 1991, for a brief review). For example, one dual-process account assumes rhar recognition decisions are based on the familiarity of the target item if that familiarity is either sufficiently strong or suf-

Parallel interactive retrieval of item and associative information from event memory

“…For them to be combined this way means that item and associative retrieval processes must yield similar outputs. This is inconsistent with theories that posit qualitatively different mechanisms for item and associative retrieval, including most dual-process theories which assume either that item and associative retrieval are independent (via separate "familiarity" and "recollection" processes Yonelinas, 1997) or that associative retrieval can only serve to suppress an item match (via "recall-toreject"; Rotello & Heit, 2000) rather than enhance it. At the same time, the ability for mismatch information to arise separately from item and associative retrieval is not easily accommodated by either single-process or summed dual-process models (Kelley & Wixted, 2001) which entail pooling of both match and mismatch information into a univariate "memory strength".…”

“…We address one final point demonstrated in prior studies of retrieval dynamics, namely, that associative retrieval is slower than item retrieval (Gronlund & Ratcliff, 1989;Rotello & Heit, 2000). From the viewpoint that associative information arises from the unitization of item information, it is clear that associative retrieval depends on having sufficient item information for a unit to be formed "on the fly" within the trial.…”

“…Where there are differences between item and associative retrieval, they are most clearly found in retrieval dynamics: People can discriminate between studied and unstudied items faster than they can distinguish learned from unlearned pair associations (Gronlund & Ratcliff, 1989;Rotello & Heit, 2000;Nobel & Shiffrin, 2001). While this delay has been attributed to an independent "recall-to-reject" process for associative retrieval (Rotello & Heit, 2000), this account predicts that partial cues will aid associative recognition (i.e., using a singly presented word to retrieve its studied associate), but in fact they do not (Gronlund & Ratcliff, 1989).…”

“…While this delay has been attributed to an independent "recall-to-reject" process for associative retrieval (Rotello & Heit, 2000), this account predicts that partial cues will aid associative recognition (i.e., using a singly presented word to retrieve its studied associate), but in fact they do not (Gronlund & Ratcliff, 1989). Paradoxically, these results are more consistent with a strong interaction between item and associative retrieval in which item information serves to "gate" associative retrieval by providing a baseline or context against which later associative information is judged (Hockley, 1991;Criss & Shiffrin, 2005;Cox & Shiffrin, in press).…”

A Dynamic Approach to Recognition Memory

“…Instead, they found that, although participants could reject novel foils (XY) more quickly when given a 200 ms head start, that same time conferred no advantage in rejecting AB' foils. Rotello and Heit (2000) contend that 200 ms is quite brief and may not have been long enough to allow for participants to use the first word as a retrieval cue, since it would take participants roughly that amount of time to even read the word. Nobel and Shiffrin (2001) directly compare the dynamics of cued recall and associative recognition in a set of signal-to-respond experiments (in recall, participants either generated a response word immediately after the signal or were allowed to make a "give up" response).…”