Academic journal article
By Perruchet, Pierre; Rey, Arnaud; Hivert, Eimeric; Pacton, Sébastien
Memory & Cognition , Vol. 34, No. 5
In an associative recognition task, distractors generally consist of a rearrangement of the items composing the study pairs. This makes it possible that processing the distractors generates retroactive interference on memory for the study pairs. In Experiment 1, we explored this possibility in a yes/no recognition test concerning previously learned arbitrary associations between visual symbols and auditory syllables. Rearranged pairs had a deleterious impact on the accuracy and the speed of responses to related correct pairs. This effect did not vary as a function of the number of training blocks, and furthermore, in Experiment 2, the same effect was observed for overlearned small multiplication facts. These results suggest that exposure to potentially confounding information generates interference even if this information is known to be incorrect. Some implications are outlined, especially with regard to the widespread use of multiple-choice tests in knowledge evaluation.
In research on associative recognition (e.g., Cleary, Curran, & Greene, 2001 ; Hockley, 1991 ; Light, Patterson, Chung, & Healy, 2004; Westerman, 2001), participants first study pairs of items (e.g., AB, CD) and then have to discriminate between intact (e.g., AB) and rearranged (e.g., AD) pairs. It is worthy of note that the rearranged pairs, or dis tractors, are patterned in a way similar to that of the interfering information in classical research on retroactive interference using paired-associate paradigms. In that context, participants are first trained with AB, then with AD. A vast amount of research has shown that processing AD has a deleterious effect on the subsequent retrieval of AB (see Anderson & Neely, 1996, for a review). This parallel between associative recognition and retroactive interference paradigms raises the possibility that processing the rearranged pairs in associative recognition tasks could be detrimental to the subsequent correct endorsement of the intact pairs. However, this possibility requires empirical confirmation, because participants in associative recognition studies are not asked to learn the AD pairs and, furthermore, they presumably appraise some of the AD pairs as incorrect. Also, the standard paired-associate paradigms require the retrieval of a response (e.g., B) given a stimulus (e.g., A), whereas a judgment on displayed pairs of events (e.g., AB) is asked for in associative recognition paradigms. Likewise, the ordering and the timing of the AB and AD events differ markedly between the two sets of studies. All of these differences make it difficult to generalize the conclusions derived from the literature on interference to associative recognition paradigms.
To the best of our knowledge, research on associative recognition has not yet explored the possibility of interference between distractor and target processing.1 The aim of the following experiments was to fulfill this objective. To do so, ayes-no recognition test, in which half of the test pairs were intact and half rearranged, was repeatedly presented. Four categories of intact AB pairs were contrasted. What gave them their specific status was that none, one, or both of their constitutive members were also part of the rearranged pairs. In Category 1, none of the members of the pairs was part of the distractors. Using a target!distractor template, these pairs will be designated hereafter as ABIXX, with X standing for an item different from A and B, respectively. These pairs were used as the baseline. In Category 2, A was sometimes paired with an incorrect partner in the distractor items. These pairs will be called ABIAX. Taken jointly, these first two categories of test items provide the materials with which to investigate retroactive interference as defined in the traditional literature. Two other categories of test items were added. In Category 3, denoted ABIXB, A was never followed by an incorrect partner, but B followed a partner other than A in some distractor items. …