An Experimental Analogue of Flashbulb Memories

This study tested the binding hypothesis: that emotional reactions trigger binding mechanisms that link an emotional event to salient contextual features such as event location, a frequently recalled aspect of naturally occurring flashbulb memories. Our emotional events were taboo words in a Stroop color-naming task, and event location was manipulated by presenting the words in different task-irrelevant screen locations. Seventy-two participants named the font color of taboo and neutral words, with instructions to ignore word meaning; in one condition, several words were location consistent (i.e., always occupied the same screen location), whereas in another condition, several colors were location consistent. Then, in a surprise recognition memory test, participants recalled the locations of location-consistent words or colors. Although attention enhanced overall location memory for colors (the attended dimension during color naming), emotion (taboo vs. neutral words) enhanced location memory for words but not colors. These results support the binding hypothesis but contradict the hypothesis that emotional events induce imagelike memories more often than nonemotional events. The present study addresses ongoing theoretical and empirical issues associated with emotionally charged, or flashbulb, memories. One issue concerns the main theoretical idea that has guided flashbulb memory research: analogies with photography and computer printouts. Several studies have demonstrated that naturally occurring flashbulb memories are less accurate or resistant to decay than these analogies would suggest (see, e.g., Christianson, 1989; McCloskey, Wible, & Cohen, 1988; Neisser & Harsch, 1992; Neisser et al., 1996). However, even when inaccurate, emotionally charged memories are experienced as extremely detailed and vivid, with imagelike or perceptual features that are unusual for ordinary memories (see, e.g., Talarico & Rubin, 2003). Perhaps nonemotional events are less likely to induce storage of perceptual images than emotional events (Livingston, 1967), even if the images are fuzzy, partial, inaccurate, and subject to decay (e.g., Pillemer, 1984). The present study tested this fuzzy-photograph hypothesis in an experimental paradigm that we argue induces analogues of flashbulb memories. Flashbulb memories have also provoked empirical controversies. The flashbulb memory concept evolved from naturalistic studies of memories for traumatic events such as the assassinations of John F. Kennedy andMartin Luther King, Jr. (Brown& Kulik, 1977), the San Francisco and Loma Prieta earthquakes (Neisser & Harsch, 1992; Neisser et al., 1996), the deaths of Princess Diana (Davidson & Glisky, 2002) and French President Francois Mitterand (Curci, Luminet, Finkenauer, & Gisle, 2001), the Hillsborough stadium disaster (Wright, 1993), the space shuttle Challenger explosion (Bohannon, 1988; Bohannon & Symons, 1992; McCloskey et al., 1988; Neisser & Harsch, 1992), the assassination attempt on Ronald Reagan (Pillemer, 1984), the resignation of British Prime Minister Margaret Thatcher (Conway et al., 1994), the verdict announcement in the O.J. Simpson trial (Schmolck, Buffalo, & Squire, 2000; Winningham, Hyman, & Dinnel, 2000), and the September 11 (2001) tragedies (Pezdek, 2002; Weaver & Krug, 2002). Most of these naturalistic studies suggest that confidence in the ability to accurately remember emotionally charged events is remarkably high, and that these events and contextual details associated with them are recalled with especially high accuracy (see, e.g., Davidson & Glisky, 2002). Examples of such contextual details are how and when participants first became aware of the event, where they were (event location), what they were doing, and who else was present (see, e.g., Bohannon, 1988; Brown & Kulik, 1977; Conway et al., 1994; Curci et al., Address correspondence to Don MacKay, Psychology Department, UCLA, Los Angeles, CA 90095-1563; e-mail: [email protected]. PSYCHOLOGICAL SCIENCE Volume 16—Number 1 25 Copyright r 2005 American Psychological Society 2001; Larsen, 1992). Although memory for contextual details may degrade over time (see, e.g., Neisser et al., 1996; but also Horn, 2001), intensity of the initially experienced emotion correlates with recall accuracy (Bohannon, 1988; Conway et al., 1994; Pillemer, 1984; Schmolck et al., 2000). However, contradictory results abound. For example, Talarico and Rubin (2003) concluded that perceived, or subjectively experienced, accuracy rather than accuracy per se distinguishes memories for the September 11 tragedies. Such conflicting results may arise because stimulus factors, such as the complexity of emotional versus nonemotional events, and attentional factors, such as importance, distinctiveness, novelty, and interest, are difficult to control in naturalistic studies (see, e.g., Cahill & McGaugh, 1995; Christianson, 1992; Christianson & Loftus, 1987, 1990, 1991; Christianson, Loftus, Hoffman, & Loftus, 1991; Larsen, 1992; McCloskey et al., 1988; Neisser & Harsch, 1992). Rehearsal and elaboration following naturalistic events are likewise difficult to control because people tend to create narrative descriptions of salient emotional events that they subsequently repeat or communicate to others (see Neisser et al., 1996; Pezdek, 2002), often many times in the weeks, months, and even years between encoding and memory test (see, e.g., McCloskey et al., 1988; Neisser & Harsch, 1992; Neisser et al., 1996; Pillemer, 1984; Weaver, 1993). As a result, flashbulb memory effects may reflect enhanced elaboration and repetitive encoding of emotional experiences during the recall interval, rather than the superior initial encoding that the term ‘‘flashbulb’’ suggests. To address these issues, we (MacKay et al., 2004) developed an experimental paradigm to assess memory for emotion-linked events and their contextual details while controlling for attention, stimulus factors, elaborative encoding, and rehearsal. The emotional events were taboo words, which enhance skin conductance, an unconscious index of emotional arousal (see, e.g., LaBar & Phelps, 1998). The contextual detail investigated was the font color of the words. Participants saw taboo and neutral words matched for length and familiarity, and named the font colors of the words as quickly as possible while ignoring word meaning. Then, in a surprise test following color naming, we tested recognition memory for the color of a subset of the taboo and neutral words that occurred in the same font color throughout the color-naming task. Like contextual details associated with naturally occurring flashbulb memories, colors were remembered better and with higher confidence ratings when associated with taboo words than when associated with neutral words. Also indicating better memory for emotion-linked context, response times on correct trials were faster for colors associated with taboo words than for colors associated with neutral words (see also Doerksen & Shimamura, 2001). However, none of these effects were due to rehearsal during the recall interval (the surprise color-recognition test followed immediately after color naming) or during the brief interval between color-naming trials (2.0 s), especially given that the words and their relations to colors were unattended and task irrelevant. Nor were these effects due to elaborative encoding (because the recognition test provided the words) or to differences in stimulus complexity, importance, or inherent interest (because color, the contextual feature tested, was counterbalanced across word type). The present study adopted similar procedures to test memory for another contextual feature in the taboo Stroop task, namely, the screen location of taboo and neutral words. Participants named the font color of words presented in different screen locations that were irrelevant to color naming. An analogue to rehearsal in the case of naturally occurring flashbulb memories was introduced by presenting some words repeatedly in the same screen location. We then tested memory for the location of these location-consistent words in a surprise recognition memory test following color naming. If word location on a monitor functions similarly to event location in naturally occurring flashbulb memories (e.g., Curci et al., 2001), then location memory should be better for taboo than for neutral words in this word-location condition. In addition, in an independent taboo Stroop condition, we manipulated colors rather than words as the location-consistent feature: Two colors always occurred in the same screen location, and different taboo words always occupied one color-consistent screen location, whereas neutral words always occupied the other. Location memory for the location-consistent colors was then assessed in a surprise memory test following color naming. The purpose of this color-location condition was to test two competing accounts of experimentally induced flashbulb memory effects. The first is the fuzzy-photograph hypothesis: that emotional reactions induce storage of taboo words and their context as perceptual images that include color, word, and location in simultaneous (but perhaps fuzzy or degraded) form. For the color-location condition, the fuzzy-photograph hypothesis predicted better memory for color in locations containing taboo words than in locations containing neutral words because emotion triggers a ‘‘now print’’ command for storing all simultaneously active information (Livingston, 1967)—in this case, color, word, and location in image form. The second account of experimentally induced flashbulb memory effects is the binding hypothesis of MacKay et al. (2004; see also MacKay, Burke, & Stewart, 1998): that emotional reactions trigger binding mechanisms that link the specific source of an emotion to salient aspects of the context in which the emotion occurs. In the case of taboo words, word meaning is the specific source of the emotion because meaning rather than orthography makes taboo words taboo. For example, a word such as ask is nonarousing and neutral in emotional tone despite sharing orthography with the taboo word ass. Under the binding hypothesis, then, word-specific emotional reactions to a taboo word would trigger binding mechanisms forming a direct and specific link between a word’s meaning and the word’s associated location in the word-location condition. However, 26 Volume 16—Number 1 Emotion, Memory, and Attention