Influence of affective 1 Running head: AFFECTIVE MEANING AND MEMORY FOR CONTEXT

In four experiments, we have investigated the influence of the affective meaning of words on memory for two kinds of contextual features that differ in the amount of effortful processes they require to be encoded in memory (i.e., color and spatial location). The main results showed that memory for color in which words were typed was better for emotional than for neutral words, but only when color information was learned incidentally. In contrast, spatial location of the words was better remembered for emotional than for neutral words whatever the encoding conditions (intentional vs. incidental). It is suggested that the influence of affective meaning on context memory may involve an automatic attraction of attention to contextual features associated with emotional words. Influence of affective 3 Influence of Affective Meaning on Memory for Contextual Information In recent years, considerable evidence has been accumulated showing that explicit memory for emotionally salient stimuli is typically better than memory for neutral stimuli (see Hamann, 2001, for review). Individuals generally tend to recall or recognize more emotional than neutral items and this memory enhancement has been reported with various kinds of materials such as films (e.g., Cahill et al., 1996), pictures (e.g., Bradley, Greenwald, Petry, & Lang, 1992; Hamann, Ely, Grafton, & Kilts, 1999; Kensinger, Brierley, Medford, Growdon, & Corkin, 2002), and words (e.g., Kensinger et al., 2002; Nagae & Moscovitch, 2002). In addition to affecting quantitative measures of memory (i.e., the amount of items that are recalled or recognized), the emotional meaning of a stimulus also seems to influence the subjective state of awareness accompanying memory. Some studies have used the “remember/know” procedure in which participants are asked to indicate whether their recognition is accompanied by a detailed sense of re-experiencing an item (in which case, a “remember” response would be made), or whether it simply seems familiar (in which case, a “know” response is made; Tulving, 1985; see Gardiner & Richardson-Klavehn, 2000 for review). It has been found that emotional words (Dewhurst & Parry, 2000) and pictures (Ochsner, 2000) were more likely to be remembered than neutral ones, which suggests that people are better able to bring back to mind some recollection of what occurred at the time an item was encoded for emotional items than for neutral items. Unfortunately, one does not know what participants actually remembered from emotional stimuli when they made a “remember” response in these studies. Memory for an item can include multiple kinds of information, such as semantic features of the stimulus, information about the time or place at which it was acquired, its modality of presentation, associated thoughts and emotions, item parameters such as size and color, and so forth (Johnson, Hashtroudi, & Lindsay, 1993). Gardiner, Ramponi, and Richardson-Klavehn (1998) Influence of affective 4 have reported data indicating that “remember” responses could reflect recollection of various kinds of information such as intra-list associations, extra-list associations, item-specific images, item’s physical features, or associated personal thoughts or memories. In addition, Perfect, Mayes, Downes, and Van Eijk (1996) showed that, when participants made a “remember” response, they were aware of one or more pieces of contextual information associated with remembered items such as their spatial location, their temporal context, or their visual appearance. The greater probability of making a “remember” response for emotional than for neutral items thus suggests that emotional stimuli are associated with a better recollection of some contextual information associated with target stimuli. However, it is currently not known whether all contextual features are better remembered when associated with emotional stimuli or whether only memory for certain kinds of contextual information is affected by the emotional meaning of stimuli. The influence of the affective meaning of the stimuli on memory for associated contextual information may indeed vary depending on the type of contextual details that are evaluated. Some contextual features are aspects of a stimulus that are inevitably processed when the stimulus is perceived and comprehended (i.e., intrinsic context), such as the case and color in which words are written, for instance. In contrast, other features are characteristics of the stimulus situation that are irrelevant to the processing of the stimulus itself (i.e., extrinsic context), such as the color of the walls of the room in which the experiment is carried out, for example (Godden & Baddeley, 1980). In addition, contextual features differ depending on whether they are automatically encoded into memory or whether they require more or less effortful processing to be encoded into memory (Hasher & Zacks, 1979). For instance, there is evidence suggesting that the spatial location of a stimulus is typically encoded automatically into memory (e.g., Andrade & Meudell, 1993), whereas its color is not (e.g., Light & Berger, 1974). When considering these distinctions among different Influence of affective 5 types of contextual information, it would be interesting to examine memory for specific contextual features (e.g., color, shape, spatial location, temporal context) separately in order to better understand the influence of affective meaning on memory. A recent study reported by Doerksen and Shimamura (2001) is relevant to this latter issue. In three experiments, the authors reported that the color in which words were typed (Experiment 1) or the color of frames that bordered the words (Experiment 2) was better remembered for emotional than for neutral words and this contextual memory enhancement did not seem to be the consequence of semantic clustering (Experiment 3). In Experiment 1, participants were presented with 32 emotional (half positive and half negative) and 32 neutral words that were typed in blue or yellow and they were asked to read each word and to remember the color in which it appeared. After a 5-min filler task, a free recall test was administered in which participants were asked to report as many words as possible without specifying the color in which they had been previously presented. Finally, participants were presented with the 64 old words interspersed with 64 new words (all words typed in black) and they were asked to decide whether a word was originally presented in yellow or in blue type, or whether it was a new word. The results first showed that emotional words were better recalled than neutral words. In contrast, item recognition accuracy (as indexed by the sensitivity parameter d’) was not significantly different for emotional and neutral words. Finally, Doerksen and Shimamura found that color memory (as determined by the proportion of old items whose associated color information was correctly identified) was significantly better for emotional than for neutral words. This experiment thus suggests that memory for some kind of contextual information (i.e., color) is better for emotional items than for neutral items. However, although color memory was significantly better for emotional than for neutral words, when one looks closer at the results, it can be argued that the performances were in fact not different from chance. Influence of affective 6 Indeed, percentages of old items whose associated color information was correctly identified were 53.5 % (± 16.0) for emotional words and 45.2% (± 15.9) for neutral words. Doerksen and Shimamura considered that chance level for color memory was 33% because participants had to choose between three possible responses (yellow, blue, or new) during the memory test. However, in this way of computing, chance levels for item recognition and color memory are tangled. Provided that participants have first to consider a word to be old before deciding in which color it was typed, there are only two possible choices for color memory (yellow or blue), which means that chance level is 50% and not 33%. When considering this issue, it appears that color memory performances were not significantly different from chance in Doerksen and Shimamura’s first study, either for emotional or neutral words, t(23) = 1.072, p > .05, and t(23) = 1.470, p > .05, respectively. Considering these ambiguities or, in any case, the subtlety of Doerksen and Shimamura’s results (in Experiment 1, the difference in color memory performance between emotional and neutral words being only 8.3%), we decided to try replicating the finding of a better memory for color information when it was associated with emotional rather than neutral words. Also, we were interested in determining whether the influence of affective meaning on memory for contextual features might extend to another important contextual information, namely spatial location. As we have already mentioned, color and spatial location are contextual features that differ in the amount of resources and attention they require to be encoded into memory. The encoding of spatial location of an item is thought to require little or no specific further attentional processing than simply attending to the item (Hasher & Zacks, 1979). In contrast, the encoding of the color of an item does require additional attentional processing (Light & Berger, 1974). Consistent with this, previous studies have found that the intention to learn contextual information improved memory for color information (e.g., Light & Berger, 1974; Park & Mason, 1982), but not memory for spatial Influence of affective 7 location (e.g., Ellis, 1990). In the present studies, we were interested in determining whether the influence of the affective meaning of words on memory for contextual information may differ depending on whether this information is encoded automatically or effortfully into memory. Accordingly, we compared the influence of affective meaning on memory for contextual information when this information was learned incidentally or intentionally. It should be noted beforehand that the present series of experiments were intended to examine the influence of the affective meaning of words and not the influence of emotional arousal. It is important to emphasize this because, although emotional words clearly have an affective meaning for individuals, they typically do not elicit strong emotional arousal as other kinds of stimuli do (see Phelps et al., 1998), so the mechanisms that influence memory in the two cases may be different (see the General Discussion for further discussion of this issue). We report two first experiments in which, contrary to Doerksen and Shimamura, we failed to find an effect of the affective meaning of words on memory for the color in which they were typed. In a third experiment, however, when eliminating a potential problem in the procedure of the two previous studies, we found that color memory was better for emotional than for neutral words, but only when color information was learned incidentally. Finally, in a fourth experiment, we found that spatial location was better remembered for emotional than for neutral words, whatever the encoding condition (intentional vs. incidental).