The Effect of Sleep on the Processing of Emotional False Memories

The Effect of Sleep on the Processing of Emotional False Memories

Chloe Newbury

2015

People often think they remember events and information that in fact never happened. In previous studies using the Deese-Roediger-McDermott (DRM) paradigm, participants viewed lists of semantically related words, and during testing were more likely to accept as seen words that were related to the lists but were actually unseen, indicating a false memory. Research suggests that sleep promotes this effect, as does the use of negatively valenced stimuli, although the effect of emotion is disputed. The current study investigated what effect emotion, in particular valence, has on false memory formation, and whether sleep promotes emotional false memories. Fifty participants were tested on their recognition performance using an emotional and neutral DRM paradigm after a 12-hour period of sleep or wake. As predicted, we found an increase in false recognition of negatively valenced lure words, as well as an overall effect of emotion, with emotional words leading to increased false recognition compared to neutral. We failed to replicate any sleep effect on performance accuracy of neutral or emotional memory, although the response time data indicates some effect of sleep on emotional memory performance. The quality of participants’ sleep and design of the current study are explored as possible explanations for this lack of a sleep effect. This study therefore indicates that emotion plays a significant role in the formation of false memories independent of sleep.

DRM
false memory

Negative and positive DRM word-lists and critical lures were taken from Brainerd, Holliday, Reyna, Yang, and Toglia (2010) who controlled for other properties that are thought to affect false memory formation, including concreteness, meaning and frequency of words (Roediger, Watson, McDermott, & Gallo, 2001). Neutral DRM lists and critical lures were taken from Stadler, Roediger, and McDermott (1999). Two separate lists were formed, one with negative and neutral words, and the other with positive and neutral words (see Appendix A for word-lists). Participants in both the positive and negative condition viewed the same five lists of neutral words, as well as ten negative or positive word-lists.
Mean valence and arousal scores for word-lists and critical lures were taken from the Affective Norms for English Words (ANEW) (Bradley & Lang, 1999). Independent samples t-tests showed that positive words had significantly higher ratings of valence than negative t(11.41) = 7.42, p < .001, and neutral words, t(13) = 7.43, p < .001. Negative words had significantly lower ratings of valence than neutral words, t(13) = 2.31, p = .038. Furthermore, negative and positive word-lists did not significantly differ in terms of arousal, t(12.92) = 0.52, p = .613, however neutral words had significantly lower ratings of arousal than positive, t(13) = 2.67, p = .019, and negative words, t(13) = 4.87, p < .001. It was also important that word-lists were controlled in terms of frequency and BAS. Frequency scores were taken from the MRC Psycholinguistic Database (Coltheart, 1981). Independent samples t-tests showed no significant difference in frequency ratings between negative and positive word-lists, t(18) = 0.18, p = .816, positive and neutral word-lists, t(13) = .35, p = .735, and negative and neutral word-lists, t(13) = 0.50, p = .624. BAS ratings were taken from the University of South Florida Free Association Norms (Nelson, McEvoy & Schreiber, 1998). There was no significant difference in ratings of negative and positive words, t(18) = 4.92, p = .629, positive and neutral words, t(13) = 0.32, p = .757, and negative and neutral words, t(13) = 0.89, p = .391. (See Appendix B for mean ratings).
For critical lures, independent samples t-tests showed that positive lure words had higher ratings of valence than negative lures, t(15.11) = 11.20, p < .001, and neutral lures, t(11) = 4.24, p = .001. Negative lures had significantly lower ratings of valence than neutral lures, t(11) = 3.62, p = .004. There was no reliable difference between ratings of arousal for negative and positive lures, t(18) = 0.22, p = .828, positive and neutral lures, t(11) = 1.08, p = .305, and negative and neutral lures, t(11) = 1.62, p = .134. There was no reliable difference between frequency ratings of negative and positive lures, t(18) = 1.14, p = .268, positive and neutral lures, t(13) = 0.55, p = .593, and negative and neutral lures, t(13) = 1.11, p = .287. (See Appendix B for mean ratings).
During testing, participants viewed 60 words in total; two previously seen from each DRM list (total of 30), the critical lure associated with each list (total of 15), and an unrelated word for each list (total of 15). Unrelated words were taken from lure words of unused DRM lists, as well as from Kousta, Vinson, and Vigliocco (2009), who developed emotional and neutral word-lists using the ANEW database. Unrelated words were matched to DRM word-lists in terms of valence, resulting in five unrelated neutral words, ten unrelated negative words and ten unrelated positive words. All words were presented in Courier new bold, black font, lower case and in 18-point.
Participants in the sleep condition were required to wear an actigraph sleep monitor to more accurately measure their time spent asleep and the number of awakenings. All participants were given a questionnaire before each session to collect data on sleep habits, caffeine and alcohol intake (see Appendix C), and those in the wake condition were instructed not to nap throughout the day.
Procedure
Participants were randomly allocated to either the wake or sleep group, with those in the wake group trained on word-lists at 9am and tested on the same day at 9pm. Those in the sleep group took part in the training session at 9pm, and were tested the following day at 9am. Participants were randomly allocated to the negative or positive stimuli condition.
During the training session, participants were first asked to fill out a questionnaire to assess sleep habits and caffeine and alcohol intake. Participants were then required to sit approximately 60cm from the computer screen, and were presented with 15 lists of 12 words presented one word at a time in the centre of the screen. They were first presented with a fixation point for 500ms before the words from one list were presented for 1500ms each. After each list participants were presented with three maths problems to solve for 1000ms each as a distractor task, in order to prevent participants from rehearsing words they had seen. Maths problems were presented in a random order for each participant, and each problem was only presented once throughout the task. After the three maths problems were presented, the fixation cross reappeared and participants were given another list to remember. The order of word-lists was randomised, and the order in which each word in a list was presented was also randomised.
Participants were then asked to return 12 hours later after a period of daytime wakefulness or overnight sleep. During the second session, participants first viewed a fixation cross for 500ms, and then the test words were presented to participants one at a time in the centre of the screen for 120ms. Participants were required to identify whether they thought they had seen the word in the previous session or not. They did this through the press of a key on the keypad, with a press of zero corresponding to an old word (previously seen), and one corresponding to a new word (previously unseen). The numbers zero and one on the keypad were labelled ‘old’ and ‘new’ respectively, to aid participants. Participants were not given a response deadline. Participants then saw the fixation point again 500ms after giving their response, before another word appeared on the screen. All words were presented in random order.

Lancaster University

data/SPSS.sav

Newbury2015

John Towse

Open

English

Data

LA1 4YF

Padraic Monaghan

MSc

Cognitive Psychology

Fifty participants (32 female, 18 male) with a mean age of 25.10 (SD = 9.25, range 18 to 62) took part in the study for course credit or as a volunteer

4-way mixed analysis of variance (ANOVA)