Abstract:

Dual-process theory holds that memory are based on two processes: familiarity and recollection. Familiarity refers to the feeling of knowing something or someone without the retrieval of additional information. In contrast, recollection includes remembering contextual information about the learning episode. While there is agreement that item memory can be supported by both processes, associative memory is generally thought to be supported by recollection only. However, recent research suggested that familiarity can also contribute to associative memory when the to-be-associated stimuli are unitized during encoding. Graf and Schacter (1989) defined this manipulation as unitization. Subsequently, a large number of behavioral studies, ERP studies, fMRI studies, and studies of older adults began to examine the effect of unitizaiton on associative memory and its processing. The results consistently showed that both top-down unitizaiton and bottom-up unitizaiton increased the contribution of familiarity to associative memory. However, some issues have been overlooked in these studies.

First, the vast majority of studies have considered only the effect of single-modality unitization on associative memory, while only a very small number of studies have compared the effect of different unitization on associative memory. A review of these studies reveals that conceptual unitization has a higher level of unitization than perceptual unitization and interactive imagery task, and that interactive imagery task may have a higher level of unitization than conceptual definition task. However, no studies directly compared the level of unitization between conceptual unitization and conceptual definition task, and between perceptual unitization and interactive imagery/conceptual definition tasks.

Second, limited by experimental material selection and memory load, most studies have not matched the level of unitization between the studied and rearranged pairs. Liu et al. (2020) first introduced the variable of unitization-congruence, and three studies consistently revealed that unitization-congruence does moderate the effect of unitization on associative memory. This finding not only help resolve the current debates in the literature concerning the influence of unitization on associative memory, but also reveal the optimal condition to benefit associative recognition from unitization—when item pairs with high level of unitization at encoding are rearranged into item pairs with lower level of unitization at retrieval.

Again, while it is widely accepted that unitization can facilitate the contribution of familiarity to associative memory, there is disagreement about the effect of unitization on the memory of the individual items that comprise the association. The ‘benefits-only’ account argues that unitization facilitates associative memory without impairing item memory, whereas the ‘benefits and costs’ account argues that unitization facilitates associative memory at the expense of item memory. A review of existing studies suggests that limited cognitive resources and semantic relatedness between the old and new words may be important in explaining these two accounts.

Finally, there are three apparently different hypotheses about the mechanisms by which unitization occurs, namely, the item hypothesis, the schema hypothesis, and the semantic elaboration hypothesis. Combining the effects of unitization on associative and item memory, the item hypothesis seems more reasonable.

In summary, we can not only compare the facilitation effect of different unitization on associative memory but also explore its lifelong development pattern in further study, provided that unitization-congruence needs to be taken into account.

Key words: unitization, associative memory, item recognition, familiarity, recollection