Abstract:

It is widely accepted that unitization can promote familiarity-based associative recognition, but its effect on recognition of individual components remains unclear. A few studies have focused on this question and shown two different accounts: One is “benefits and costs” account which argues that unitization promote associative recognition at the cost of item recognition, the other is “benefits-only” account which holds that unitization can promote associative recognition without impairing item recognition. In the current study, we aimed to explore how unitization influence associative and item recognition.
Twenty-nine participants took part in the study. To avoid fatigue effects, three study-test cycles were completed with a short break (2 min) in between. For each cycle, 96 word pairs were encoded at a rate of 4 s each, with a 900~1100 ms fixation cross between trials. Forty-eight word pairs were presented in compound word pairs (CW) and 48 word pairs were presented in non-compound word pairs (NCW). After a 2-min distracting phase, participants took part in an associative recognition test, in which 64 word pairs were presented: (1) CW-intact word pairs, (2) CW-rearranged word pairs, (3) NCW-intact word pairs, and (4) NCW-rearranged word pairs. In order to matched the level of unitzation between the studied and tested word pairs, two compound word pairs were rearranged into a new compound word pairs, and the same is true of non-compound word pairs. The remaining four words were used as old stimuli in item recognition test. After all three cycles are completed, participants then took part in an item recognition test. The item test was also divided into three cycles. For each cycle, 96 single words were presented: (1) Compound-old words, (2) Non- compound-old words, and (3) new words. In both associative and item recognition tests, participants were instructed to press the “F” if the word pairs or words had been learned at encoding and to press the “J” otherwise. Meanwhile, the EEG was recorded.
First, the results showed higher level of unitization [t(28) = 44.50, p < 0.001, Cohen’s d = 8.26] and faster RTs [t(28) = −6.44, p < 0.001, Cohen’s d = −1.19] for compound word pairs than for non-compound word pairs at encoding. It indicated that the manipulation of unitization was effective in the current study. Second, concerning associative recognition, an enhanced recognition performance [t(28) = 5.67, p < 0.006, Cohen’s d = 1.05], with a larger familiarity-related FN400 effect (p = 0.015) and recollection-related LPC effect (p = 0.035), was observed for compound word pairs than for non-compound word pairs. This results suggested that unitization could improve associative recognition performance through increasing the contribution of familiarity and recollection simultaneously. And finally, an equivalent item recognition performance [t(28) = 0.97, p = 0.34] between the two word pairs was found, despite the compound word pairs elicited a larger FN400 effect than the non-compound word pairs (p = 0.049). This indicated that unitization did not impair the item recognition performance (see Fig. 1).
In summary, the current study suggests that unitization not only facilitates associative recognition but also does not impair item recognition, supporting the “benefits-only” account. Importantly, familiarity can support associative recognition when the two items were unitized into a new presentation. This means that unitization is an effective strategy for improving associative memory, especially for groups with impaired recollection.

Key words: unitization, associative recognition, item recognition, FN400 effect, LPC effect