Otemon Gakuin University
Summary.  This study examined the relationship between the recall of chosen words and congruity between the selection criterion and type of pairs in intentional memory. Participants were asked to select a more pleasant or unpleasant word from two types of pairs, pleasant–neutral and unpleasant–neutral, and to remember the chosen word presented in the learning phase, followed by the free recall phase. Although the pleasant words selected from the pleasant–neutral pairs were equally recalled as in the unpleasant–neutral pairs, the chosen unpleasant words from unpleasant–neutral pairs were recalled more often than those in the pleasant–neutral pairs. These results suggest that the recall of the chosen words was determined by the congruity between the criteria of choosing and the type of paired words and the negativity bias in memory.
Target items to be remembered among alternative items chosen by participants in the self-choice condition were recalled more often than those chosen by the experimenter in the forced-choice condition). This is called the self-choice effect (Takahashi, 1989), and many hypotheses are proposed to explain this effect. The early research on self-choice effects were conducted by Perlmuter, Monty, and colleagues (Monty et al., 1973, 1979; Perlmuter & Monty, 1982). They proposed the motivation hypothesis (Perlmuter & Eads, 1998). According to this hypothesis, choosing a word leads to higher motivation among the participants, leading to a higher memory performance level, and explains the self-choice effect. However, this hypothesis had a critical problem; the motivation index was not clarified. Hence, an increased motivation level to facilitate memory performance in the self-chose condition could not be demonstrated. According to Toyota (2013), since the motivation hypothesis, others have been proposed, including the metamemory hypothesis proposed by Takahashi (1991), the multiple-cue hypothesis by Watanabe (2001), and connective processing by Hirano & Ukita (2003). However, these did not sufficiently explain the self-choice effect because rebutting evidence was observed on them.
In contrast, Toyota et al. (2007) proposed the integration hypothesis. In their study, participants were asked to choose one word they liked more in each word pair in a self-choice condition, whereas one of the paired words was underlined to signify that it was favored by another participant in a forced-choice condition. Participants were required to indicate whether their judgment differed from the other participants. Both a self-choice and forced-choice conditions were followed by unexpected free recall tests. The self-choice effect (the superiority of self-choice to forced-choice conditions) was apparent only for the pleasant–unpleasant pair. In the pleasant–unpleasant pair, a contrast between words in each pair was apparent; hence, participants could choose a word by referring to the contrast between the two words in each pair. However, in the pleasant-pleasant and unpleasant-unpleasant combination, there was no apparent contrast between the two words in each pair. These results were interpreted as follows. The pleasant–unpleasant pairs gave the participants an apparent criterion for selecting one of each pair because the pair had an opposite value; they could easily choose the word they liked more. Chosen words were integrated into participants’ cognitive structure and heightened precisely and would be recalled more. In contrast, the criterion for choosing was not apparent for the pleasant–pleasant and unpleasant–unpleasant pairs because each word in a pair had a similar value. Participants could not easily choose one word based on likableness (pleasantness) or dislikeableness (unpleasantness). Thus, Toyota et al. (2007) considered that an apparent criterion would support the participant to integrate each chosen word into a cognitive structure to facilitate their recall. Toyota and Kobayashi (2009), using an intentional memory procedure, also found a self-choice effect only in pleasant–unpleasant pairs. This finding also supports the integration hypothesis. Initially, the word “integration” in memory was used by Craik and Tulving (1975). They found that the words in the congruous contexts were recalled or recognized more often than those in incongruous contexts; the superiority of congruous contexts is called the “congruity effect” and has been observed in several studies (Goldman & Pellegrino, 1977; Hall & Geis, 1980; Toyota, 1996). According to Craik and Tulving (1975), words in the congruous contexts were integrated into the cognitive structure, but words in incongruous contexts were not. The difference in integration led to a difference in recall performance between congruous and incongruous contexts (Toyota, 2013).
Toyota et al. (2007) considered the self-choice effect using this concept of integration and suggested it is caused by the degree of integration when a word is chosen based on the participant’s criterion. For example, in a pleasant–unpleasant pair, the chosen words can easily integrate into a cognitive structure because the participants could quickly choose a word based on the contrast between each word in a pair. However, in the pleasant–pleasant and the unpleasant–unpleasant pairs, the chosen words could not easily integrate into the cognitive structure because the participants could not easily choose a word based on the contrast between each word in a pair. The selection criterion would be critical if the integration function explained the self-choice effect. Namely, the clear instruction about the criterion for choosing a word may lead to the self-choice effect in the pleasant–pleasant and unpleasant–unpleasant pairs, wherein there is no contrast between each word in a pair. Toyota (2013) used specific instruction about the selection criterion and found the self-choice effects in both types of pairs (pleasant-pleasant and unpleasant-unpleasant). Although the self-choice effects were observed in both types of pairs, the significance of its effect differed: when the participants were asked to choose the more liked word, the self-choice effect was more significant in pleasant-pleasant pairs than in unpleasant-unpleasant pairs, whereas when asked to choose the more disliked word, it was more significant in unpleasant-unpleasant pairs than in pleasant-pleasant pairs. Thus, the self-choice effects in the congruous contexts were more significant than in incongruous contexts. This suggests that the congruity between the criterion of choosing (e.g., pleasantness) the word from the alternatives and types of paired words (e.g., pleasant-pleasant pairs) is critical. However, in Toyota’s (2013) study, the participants chose words with a similar value of pleasantness or unpleasantness, making it difficult for each participant to choose the more pleasant or unpleasant words. According to Toyota (1997), many studies have indicated the effectiveness of encoding difficulty on memory performance (Auble & Franks, 1978; Einstein, 1976; Ellis, Thomas & Rodrguez, 1984; Jacoby, 1978; Kitao & Kaneko, 1981; Krinsky & Nelson, 1981). Toyota (1997) indicated that difficulty in choosing a word to memorize in each word pair facilitated the retention of chosen and non-chosen words, called the encoding difficulty effect. Although some hypotheses explained this effect, including mental effort (Tyler, Hertel, McCallum & Ellis, 1979), elaboration (Craik & Tulving, 1975), and distinctiveness (Jacoby, Craik & Begg, 1979), Kitao and Kaneko (1981) explained that such difficulty led to increased effort in elaborating the word, leading to cues for retrieving the chosen words for elaboration and distinctiveness (Toyota, 1997). This study used the two types of word pairs, pleasant-neutral and unpleasant-neutral pairs, including neutral words, to examine the effect of congruity on retention eliminating the encoding difficulty effect.
The first hypothesis of this study is as follows. Suppose the congruity between the criterion and the word type determined the recall performance of the chosen word, then: if participants were asked to choose the more pleasant word, the chosen words in pleasant–neutral pairs would be recalled more often than those in unpleasant–neutral pairs. In contrast, if asked to choose the more unpleasant word, the chosen words in unpleasant–neutral pairs would be recalled more often than those in pleasant–neutral pairs.
Baumeister et al. (2001) reviewed several psychological phenomena and proposed the focusing notion of “negativity bias,” wherein negative emotions or memories were recalled better than positive ones. Specifically, unpleasant episodes elicited more robust emotional encoding than pleasant episodes. Thus, target words associated with an unpleasant mood are processed more deeply and intensely than those associated with a pleasant mood. Toyota (2014) examined negativity bias using an incidental free recall procedure, and participants with a low ability to manage and regulate emotions recalled targets associated with unpleasant episodes more often than those with pleasant episodes. Although only for low ability participants, a negativity bias was observed in incidental memory in Toyota’s (2014) study. If the negativity bias was robust, the chosen unpleasant words in unpleasant–neutral pairs were predicted to be recalled more often than the chosen pleasant words in pleasant–neutral pairs. The second aim of this study is to examine this hypothesis.
METHODS