This is a term paper I wrote for a course in Research Methods at the University of Southern California, Fall 2011.
—————————————————————————————————————————————————————————
Effects of Residual Emotion on Memory
Frank Kotsianas
University of Southern California
kjbkjh
Author Note
This research report summarizes a semester-length student assignment in a Research Methods course. The report depicts an imagined investigation by the author. It is a “thought experiment.” The instructor for this course, William Breland, Ph.D., has simulated the data without regard for any particular student’s hypothesis. Therefore, the results should not be interpreted as a true test of the theoretical relationships expressed by the author. Furthermore, the 3×2 factorial design is a constraint that has been imposed on the author by the instructor – the design is not necessarily one that the student would propose to test the relationships of her/his own volition.
space
ABSTRACT
A large body of evidence has shown that emotion plays a role in determining the strength of a memory. However, relatively little is known about the circumstances under which emotion aids memory. In this study, we investigated whether concurrent “disembodied” emotional activation, unrelated to the stimulus, is sufficient to improve memory. Using an anger-inducing stimulus, we manipulated level of arousal in men and women and tested recall and recognition of words using the Rey Auditory Verbal Learning Test. Significant results were found for low levels of arousal for both genders, as well as significant interactions of level of arousal and gender. Together, these results indicate that “disembodied” emotional arousal is sufficient to improve memory, with the optimal amount varying by gender.
INTRODUCTION
Historically, research on emotion and memory can be found as far back as 1978. It is well documented that emotion raises alertness and so enhances recall of an item (Posner 1978, from Strongman & Russell, 1986), which may play into the correlation between anterior cingulate activation (which plays in various functions involving attention) and awareness of emotions (Posner & Rothbart, 2000). In the particular domain of lexical memory, it has been demonstrated that recall of emotional words is greater than recall of nonemotional words (Davidson, Luo, & Burden, 2001), that the application of emotional labels to words increases recall (ibid.), and that words that arouse stronger emotion are easier to recall (Bock & Klinger, 1986). In contrast, it has been shown that words that are easier to represent in imagery (ibid.), regardless of strength of emotion that the word evokes, are easier to recall. This indicates that emotion is important in recall, but cannot stand on its own as the only important precursor to memory. Multimodal processing also plays a strong role.
The goal of this study is to find out under what circumstances emotion plays a strong role in memory formation. If emotion is indeed the “rudder” that guides attention and cognition (Immordino-Yang & Damasio, 2007) and acts as a “marker” for important information (Dunn et al., 2006; Damasio et al., 1990), then can our brains be “fooled” into placing additional weight on a piece of information by emotional arousal that is unrelated to the stimulus itself? Can we “wrongly” encode information because our emotions tell us it is useful? If so, can learners’ anxiety, anger, or fear be turned to help them learn?
Formally, our question is whether any emotional processing at all affects recall and recognition of information, or if the emotion must be directly related to the item trying to be learned. The current findings indicate that the words must have emotional content or be emotionally relevant to the learner, but it remains uninvestigated if concurrent emotional activation is enough to increase memory for the words. More succinctly, is recall of words dependent on the words being inherently emotional, or on the concurrent experience of emotion? Furthermore, is the effect of emotion different for men and women?
Isen and Shalker (1982) demonstrated that participants with artificially-invoked positive emotion judged stimuli more positively. This would seem to follow from a Hebbian “fire-together, wire-together” perspective. If unrelated emotional processing has an effect on decision-making and evaluative processes, then it seems natural that concurrent activation of emotion and stimulus will be sufficient to increase recall as well. In contrast, the theories of state-specific processing (Godden & Baddeley, 1975, 1980; Goodwin et al., 1969; Overton, 1968), its subcategory, language-specific processing (Marian et al., 2007), and resource-allocation models of memory (Loftus et al., 1987), which propose that emotions “eat up” resources and so encode only information relevant to the stimulus, contend that artificially-evoked emotion should have little effect on memory. Here, we hypothesize that concurrent activation of emotion unrelated to the stimulus (“disembodied” emotion) will be sufficient to increase recall and recognition of words, and that this effect will be greater for women than for men.
METHODS
The hypothesis was investigated by the evoking of emotional arousal through a frustrating puzzle game, with measurements of arousal by the Emotional Reactivity Scale and Galvanic Skin Response (GSR), and measurement of recall and recognition of words as assessed by the Rey Auditory Verbal Learning Test (RAVLT). The effects of gender and emotional arousal on recall were assessed.
Participants
The sample included 300 individuals recruited from the University of Southern California psychology research pool and general student population. 50% were women (N = 150) and 50% men (N = 150). All participants were between 18 and 30 years of age, were primarily of White descent, and identified as members of the Western culture. They were of varied socioeconomic status (SES), though biased towards the upper middle-class. Participants were excluded from participation if they reported repeated occurrence or clinical diagnosis of extreme depression, mania, or anxiety, out of concern for extreme responsiveness to emotions.
Instruments
Emotional Reactivity Scale (ERS). The Emotional Reactivity Scale is a 21-item self-report questionnaire that measures emotion sensitivity, intensity, and persistence (Nock et al., 2007). As traits such as anxiety, depression, and mania have been shown to be mediators of memory (Williams et al., 1997), we believe that it is important to control for such conditions. The scale will be administered by the experimenter and used to identify any subjects in the extremes, who may have conditions of mania, depression, or anxiety that are not applicable to our population of interest. Participants who score particularly high or particularly low on the ERS will be excluded from analysis, and their results will be used for analysis in a further paper. Scores will also be correlated with psychophysiological data to investigate the expected correlation between ERS scores and physiological response data.
Responses to the ERS occur on a 5-point Likert scale ranging from 0 (“Not at all like me”) to 4 (“Completely like me”). The overall score will be calculated by averaging scores across all items, with a higher score indicating higher reactivity to and intensity of emotions, and later used to normalize skin conductance scores to yield one overall measure of emotional arousal. Prior use of this scale with patients with severe depression indicated high internal consistency (Cronbach’s a = .94) (Nock et al., 2007). Similar consistency is expected in the current sample.
Skin Conductance. It has been demonstrated that skin conductance change accompanies emotional response due to change in the amount of moisture secreted by the sweat glands of the hand (Martini & Bartholomew, 2003), and that size of skin conductance response to an item is positively correlated to later recall of that item (Corteen, 1969), giving some evidence for an interaction between skin conductance and memory. Recordings will be taken using a standard galvanometer attached to the index and middle fingers of the primary hand. Measurements will be taken during presentation of the emotion-invoking stimulus and during presentation, recall, and recognition of word lists in the RAVLT.
Rey Auditory Verbal Learning Test (RAVLT). The RAVLT is a widely-used clinical evaluation of many memory functions, including recall, recognition, retrieval, encoding, storage, and interference of words. Here, we are most interested in the assessment of recall and recognition. The test uses two lists of 15 words, referred to as List A and List B, spoken by the experimenter at the rate of 1 per second.
The procedure is as follows: List A is presented, and the subject is asked to repeat as many words as he or she can remember, in any order. This occurs a total of five times. Then List B is presented, and the subject is asked to repeat as many words as he or she can remember, in any order. Then, the subject is asked to recall as many words of List A as he or she can. Then there is a 30 minute delay, followed by 1 trial of recall of List A. Recognition is then tested by reading 30 words and asking which of the 30 were previously read from List A.
Recall and recognition scores for the recall trials will be collapsed to create one composite measure of the percentage of words recalled. Recognition score will similarly be converted to a percentage of words recognized, so as to provide an intuitive basis for comparison. Reliability for the RAVLT has been reported to be 0.70 for List A, and 0.38 for List B (Snow, 1988).
Stimulus. The emotion-inducing stimulus will be a puzzle game played on a computer whereby it is impossible to win. Whatever response the player chooses, the computer displays the opposite answer as correct. To increase pressure and propensity to feel anger, subjects will be told inciting remarks at successive intervals: for example, that they must complete the puzzle in under five minutes; that others have done it in under three minutes; that others have done it faster than they; and that “the answer is really obvious!”
In pilot testing, we confirmed that this instrument reliably creates high and low levels of arousal in our population, as measured by GSR, and can differentiate between the two levels.
Procedure
Participants will be randomly selected from the USC psychology research pool. They were informed of the purposes of the study, and all give consent to participate. Participants reporting diagnoses or recurring experiences of depression, anxiety, or mania are excluded from the study.
Participants will be administered the Emotional Reactivity Scale by either the experimenter or a trained research assistant. Participants who score extremely high or extremely low on the scale (indicating possible depression or mania, and so not applicable to our subject population) will be excluded from analysis. Results will still be obtained for purpose of investigating the existence of a significant difference from more average scorers.
Participants will be randomly assigned to one of 3 manipulated conditions: high arousal, some arousal, or no arousal. Those who show higher or lower levels of arousal compared to baseline, than accounted for by their condition, will be excluded from analysis.
On the day of the study, participants will come into the laboratory, and after introduction and signing of paperwork, will be connected to the Galvanic Skin Response device. A baseline measurement will be taken for purposes of normalization and comparison with results in “emotional arousal” conditions. These measurements will continue to be taken throughout the computer puzzle stimulus and the administration of the RAVLT.
Participants in the “emotional arousal” conditions will be instructed that they will play a puzzle game. Measurements of GSR will be taken while they play the game. This game is impossible to win. Any response the subject gives is judged incorrect by the computer, which induces frustration and anger in the participant. To increase frustration, pressure, and propensity to feel anger, subjects will be told at successive intervals that:
- they must complete the puzzle in under 5 minutes,
- that others have done it in under 3 minutes,
- that others have done it faster than they, and
- asked, “You still didn’t get it?”,
- Possibly followed by a comment such as, “But it’s so obvious…”.
The comments shall be distributed at 2-minute intervals, until the subject has reached the arousal level indicated by his treatment condition (High or Low).
Participants in the “no arousal” condition will play a game exactly similar in appearance, only this one is not rigged. This is so as to control for any effects of puzzle-solving or game-play. There may be some effects of the game on emotion (for example, joy at succeeding), but we count this as noise attributable to individual differences, such as one’s puzzle-solving skill. We trust that this will be controlled for in analysis.
After the game is played, the experimenter will enter and administer the Rey Auditory Verbal Learning Test. He will track results for recall of List A and List B, for delayed recall of List A, and for recognition of List A. GSR will be taken here for purposes of noting abnormal deviations from previous levels, though none is expected. Subjects in all conditions will participate in this step of the study.
The subject will be debriefed as to the purpose of the study, the experimenter will assess their state to ensure that they have been restored to their previous state after the testing procedure, compensated, and thanked for his or her participation.
ANALYSES
A three-way between-subjects MANCOVA, two univariate analyses of covariance (ANCOVAs), and post-hoc analyses using a Bonferroni adjustment will be conducted to assess for significant differences between the high-arousal, low-arousal, and no-arousal conditions, between males and females, and for interaction effects between the two independent variables.
One will be conducted for recall data, and one for recognition data. We expect to see an effect of high arousal, an effect of low arousal, and a possible interaction of high arousal and gender-female. We will also measure three covariates: age, socioeconomic status (SES), and IQ.
There are a few relevant data collapses and transformations of data. Scores for Recall are a collapsed measure of the participant’s recall of List A and List B of the RAVLT. Scores for Recognition are a measure of the participant’s recognition of List A words as assessed by the RAVLT. All classifications of arousal level are determined by the subject’s difference between baseline, assessed before any task, and average level of arousal during the stimulus computer game.
As mentioned before, subjects who exhibit levels of arousal that exceed the boundaries of their experimental condition will be excluded from analysis.
RESULTS
A 3×2 multivariate analysis of covariance (MANCOVA) was conducted to determine the effects of three levels of evoked arousal (low, high, and no arousal) and gender (male and female) on words recalled and words recognized in the Rey Auditory-Verbal Learning Test (RAVLT). Significant effects were found for the level of evoked arousal on the two dependent measures, Wilks’ lambda = .856, F(4, 580) = 11.69, p<.001. Significant effects of the interaction between level of evoked arousal and gender on the dependent measures of memory were also found, Wilks’ lambda = .872, F(4,580) = 10.29, p<.001. No significant effects were found for the effect of gender alone (p>.05). Tables 1 and 2 report the means and standard deviations of each dependent variable for the six emotional-arousal-level-by-gender groups.
Two univariate analyses of covariance (ANCOVAs) were conducted for words recalled and words recognized to assess the effect of emotional arousal and the interaction of emotional arousal and gender on each dependent variable. Specific group comparisons were conducted using t-tests with the Bonferroni method of adjusting a to control for increased Type I errors after multiple comparisons. The ANCOVA regarding emotional arousal level effects on words recognized was significant, F(2,291) = 19.71, p<.001; however, the ANCOVA regarding emotional arousal effects on words recalled was not (p>.05). With respect to the interaction between level of emotional arousal and gender, effects were significant for both words recalled, F(2,291) = 8.074, p<.001 , and for words recognized, F(2,291) = 12.707, p<.001. No significant main effects of gender on either words recalled or words recognized were observed (p>.05 on both counts).
Post-hoc analyses were conducted to assess the significant differences observed between levels of emotional arousal on words recognized and between the six arousal-by-gender groups on both words recalled and words recognized. These analyses were conducted by examining the Bonferroni-adjusted confidence intervals for difference from zero. Table 3 presents the 95% confidence intervals around the group means for the significant effect of level of emotional arousal observed on words recognized, and Tables 4 and 5 present the 95% confidence intervals around the group means for the significant interaction of level of emotional arousal and gender on words recalled and words recognized.
With respect to words recalled, we investigated the interaction between level of emotional arousal and gender (Table 4). While females showed no significant differences between words recalled at any level of evoked, unrelated emotional arousal, males recalled significantly more words under conditions of high arousal (95% CI from .201 to .726 and 95% CI from -.618 to -.093, respectively). Though words recalled under varying levels of arousal did not differ significantly from each other for females, there was an interesting trend showing that women recalled more words under conditions of low arousal than under conditions of high arousal, and marginally more words under conditions of no arousal than conditions of high arousal (see Figure 1). Though none of these differences were significant in our study, this trend may bear further investigation.
With respect to words recognized, we investigated the effect of level of evoked arousal and the interaction effect of level of arousal and gender (Tables 3 and 5). Across both genders, low levels of arousal resulted in significantly more words recognized than did levels of high arousal and no arousal (95% CI from .278 to .890 for low compared to high arousal, and 95% CI from .457 to 1.069 for low compared to no arousal). Interestingly, high arousal did not significantly differ from no arousal in the number of words recognized across genders (95% CI from -.127 to .484 from high to no arousal). Between specific genders, males recognized significantly more words under conditions of low arousal (95% CI from .473 to .973), and significantly fewer words under conditions of no arousal (95% CI from -.903 to -.403). Females demonstrated significantly worse recognition under conditions of high arousal than under either low or no arousal (95% CI from -.512 to -.012). For illustration, see Figure 2.
DISCUSSION
In this study, we investigated the effects of emotion unrelated to a stimulus (“disembodied” or “unrelated” emotion) on memory by looking at levels of disembodied arousal and the number of words participants were able to remember. Specifically, we evoked disembodied arousal using a frustrating stimulus and measured the number or words participants recalled or recognized on a subsequent Rey Auditory Verbal Learning Test (RAVLT).
We found higher that high levels of unrelated arousal predicted greater recall of words in males, and that no arousal predicted significantly worse recall of words for males. In recognition, we found that males recognized significantly more words under conditions of low arousal and significantly fewer words under conditions of no arousal. Females demonstrated significantly worse recognition under high arousal than under either low or no arousal. When considered together, we found that low levels of disembodied arousal helped recognize significantly more words than high or no arousal.
This seems to indicate that disembodied arousal can cause males to remember words better, though the amount that aids memory varies with the type of remembering desired. In contrast, disembodied arousal seems to adversely affect female memory when experienced in large doses.
Overall, this seems to provide evidence for three conclusions: that higher levels of arousal are required to recall words than to recognize them; that some arousal is helpful for memory for words, but high levels of arousal can interfere with memory; and that males and females respond differently to equal levels of emotion. The first conclusion is consistent with prior findings about recall and recognition. The second conclusion also seems to correlate with the intuition that too much emotion would “override” the senses, and would result in fewer words remembered. The third conclusion adds to the second.
The most interesting result of our experiment was that males and females respond to disembodied emotion differently. It seems that men require high arousal to recall words, and only low arousal to recognize them. In fact, arousal beyond low levels inhibits word recognition in men. In women, words recalled and words recognized are approximately the same, though high arousal inhibits word recognition but not word recall (Figures 1 & 2). This would seem to indicate that recall is more difficult than recognition, a well-known result, and that women are not only less affected than men by low levels of disembodied emotion, but are more affected than men by high levels of disembodied emotion; while men can be “convinced” to recall something by high levels of emotion or to recognize something by low levels of evoked emotion, women are unresponsive to low levels of unrelated emotion and get worse with high levels of unrelated emotion.
We can think of two plausible explanations to account for this finding: that women require emotions to be tied directly to the stimulus to aid in memory, and become overwhelmed by high levels of arousal that are unrelated to a stimulus, explaining why they respond less than men to equal levels of arousal and become worse under high levels of arousal; or that women are more aware of the sources of their emotions than men, and so men can be “convinced” by unrelated emotional activity to recall and recognize given stimuli, while women are able to ignore this deception until it becomes too great, at which point they actively inhibit encoding of the stimulus (as demonstrated by the decrease in recognition under high evoked arousal).
Given the theory that emotion is an indicator of importance and relevance to an individual’s interests (Immordino-Yang & Damasio, 2007), these results indicate that women have a better grasp of what is a relevant emotion and what is not. This may be because our culture encourages women more than men to be “in-touch” with and to sort through their emotions, so women may be better able to differentiate between and understand their emotions than men.
Given our goal of understanding disembodied emotion in the classroom, it would seem that evoked emotion would aid male students in recalling and recognizing information, but could actually hurt the performance of female students. This would indicate that evoked disembodied emotion, in the vein of “pumping students up”, is not a helpful practice for learning, and would preclude it from being used as a strategy in all but single-sex male-only schools.
Limitations and Strengths
The results of the study may have been biased by the population used and the methodologies and instruments used to gather data. The population was drawn from a pool of USC undergraduates, who are primarily between the ages of 18 and 22, predominantly white, and middle- to upper-middle class. Their mnemonic ability may not be comparable to older populations or populations from different cultural or socioeconomic backgrounds. We have limited our population of interest to adults between the ages of 18 and 30, but that may not be narrow enough.
As the study tested a new paradigm and a new method of reinforcing memories (through unrelated emotional processing), it is unknown what factors may confound the results we have collected. There may be an error with our frustration paradigm, where the evoked emotion is due to factors other than the game being played. In addition, we measured emotional arousal through Galvanic Skin Response (GSR), which does not capture all the dimensions of emotional arousal. There may be some confound of emotional arousal which was not detected by our instruments. Finally, we evoked only one type of emotion (frustration). The findings for the emotion of frustration may not generalize to other emotions. There may be effects for disembodied frustration that do not exist for embodied frustration, and effects of disembodied sadness that do not exist for disembodied frustration. In addition, it can be argued that frustration is not a basic emotion in the style of Ekman (1972, 1999), but rather a composite emotion of, for example, anger and excitement, or anger and contempt. In light of that, our results are quite narrow in scope, although the specific effects found are significant. A similar paradigm could be used to test other emotions for effects.
Our study had advantages over previous studies because of its large sample size and discrimination among experimental conditions. With a sample size of 300, our error is low and our power is high, which lets us claim our findings for males and females with a high degree of confidence. We also defined our experimental classification of high, low, or no emotion according to standard deviations from individual norms of GSR, which implicitly controls for individual differences in arousal. We further controlled for extreme responses to emotional stimuli by screening participants using the Emotional Reactivity Scale and excluding individuals who reported conditions of chronic anxiety, mania, or depression from participation. Such controls increase the validity of our results.
Future Directions
It remains to be seen how disembodied emotion and stimulus-relevant embodied emotion differ in ability to affect recall and recognition of words. Though this study established that disembodied emotion does produce an effect on memory, and an effect that differs between the sexes, it is unclear how strong that effect is compared to the effects of naturally-arising emotion. It is also unknown how the effects of disembodied emotion endure or decay over time, and whether this follows a different trajectory than naturally-formed memories. Lastly, the interactions of disembodied emotion with state-specific recall remain to be seen. In this study, we tested memory directly following evocation of the emotion and presentation of the stimulus. The decay of this artificially-reinforced memory over time, and the amount of its effect that is due to being in the same emotional state, remain to be investigated. It seems likely that disembodied memories would fade more quickly over time, and would be less accessible when not experiencing the same emotional state in which they were created, but these hypotheses would have to be tested by further experiments. Such studies should also include comparisons with normally-formed emotional memories. If an advantage of disembodied memory over embodied memory is found, it could have great impact on learning and training, and could aid in understanding the mechanism behind state-specific memories.
Tables & Figures
Table 1 – Means and Standard Deviations for Words Recalled on RAVLT across Level of Emotional Arousal and Gender.
Words Recalled
|
Main Effects |
Interaction Effects |
Means |
Standard Deviations |
| Gender – Male |
-.039 |
1.150 |
|
| Gender – Female |
.039 |
.825 |
|
| Arousal – Low |
-.001 |
.869 |
|
| Male x Low Arousal |
-.224 |
.692 |
|
| Female x Low Arousal |
.221 |
.972 |
|
| Arousal – High |
.194 |
1.017 |
|
| Male x High Arousal |
.463 |
1.209 |
|
| Female x High Arousal |
-.075 |
.693 |
|
| Arousal – None |
-.193 |
1.075 |
|
| Male x No Arousal |
-.356 |
1.299 |
|
| Female x No Arousal |
-.029 |
.769 |
mjbgjhg
Table 2 – Means and Standard Deviations for Words Recognized on RAVLT across Level of Emotional Arousal and Gender.
Words Recognized
|
Main Effects |
Interaction Effects |
Means |
Standard Deviations |
| Gender – Male |
.020 |
.945 |
|
| Gender – Female |
-.020 |
1.055 |
|
| Arousal – Low |
.450 |
.943 |
|
| Male x Low Arousal |
.722 |
.619 |
|
| Female x Low Arousal |
.178 |
1.124 |
|
| Arousal – High |
-.136 |
.905 |
|
| Male x High Arousal |
-.008 |
.942 |
|
| Female x High Arousal |
-.263 |
.857 |
|
| Arousal – None |
-.314 |
.994 |
|
| Male x No Arousal |
-.654 |
.692 |
|
| Female x No Arousal |
.025 |
1.132 |
Table 3 – Confidence Intervals (CIs) for Significant Effects of Level of Emotional Arousal on Words Recognized
Words Recognized
| Level of Arousal I | Level of Arousal II | Lower Bound of CI | Upper Bound of CI | Significantly Different from 0? |
| Low | High |
.278 |
.890 |
Yes |
| None |
.457 |
1.069 |
Yes |
|
| High | Low |
-.890 |
-.278 |
Yes |
| None |
-.127 |
.484 |
No |
|
| None | Low |
-1.069 |
-.457 |
Yes |
| High |
-.484 |
.127 |
No |
dfsdgfsdg
Table 4 – Confidence Intervals for Significant Effects of Level of Emotional Arousal and Gender on Words Recalled
Words Recalled
| Level of Arousal | Gender | Lower Bound of CI | Upper Bound of CI | Significantly Different from 0? |
| Low | Male |
-.486 |
.039 |
No |
| Female |
-.044 |
.481 |
No |
|
| High | Male |
.201 |
.726 |
Yes |
| Female |
-.337 |
.188 |
No |
|
| None | Male |
-.618 |
-.093 |
Yes |
| Female |
-.291 |
.234 |
No |
space
Table 5 – Confidence Intervals for Significant Effects of Level of Emotional Arousal and Gender on Words Recognized
Words Recognized
| Level of Arousal | Gender | Lower Bound of CI | Upper Bound of CI | Significantly Different from 0? |
| Low | Male |
.473 |
.973 |
Yes |
| Female |
-.075 |
.425 |
No |
|
| High | Male |
-.258 |
.242 |
No |
| Female |
-.512 |
-.012 |
Yes |
|
| None | Male |
-.903 |
-.403 |
Yes |
| Female |
-.224 |
.276 |
No |
space
Figure 1 – Trends of Interaction Between Level of Emotional Arousal and Gender on Words Recalled
Figure 2 – Trends of Interaction Between Level of Emotional Arousal and Gender on Words Recognized
space
REFERENCES
Bock, M., & Klinger, E. (1986). Interaction of emotion and cognition in word recall. Psychological Research, 48, 99-106.
Corteen, R.S. (1969). Skin Conductance Changes and Word Recall. British Journal of Psychology, 60: 81–84.
Damasio, A.R., Tranel, D. & Damasio, H. (1990). Individuals with sociopathic behavior caused by frontal damage fail to respond autonomically to social stimuli. Behavioural Brain Research, 41, pp. 81–94.
Davidson, D., Luo, Z. & Burden, M. (2001). Children’s recall of emotional behaviors, emotional labels, and nonemotional behaviors: Does emotion enhance memory? Cognition and Emotion, 15(1), 1-26.
Dunn, B. D. Dalgleish, T. & Lawrence, A. D., 2006. The somatic marker hypothesis: A critical evaluation. Neuroscience and Biobehavioral Reviews, 30, pp. 239–271.
Ekman, P. (1972). Universals and Cultural Differences in Facial Expression of Emotion. In J. Cole ed. Nebraska Symposium on Motivation. Lincoln, Nebraska: University of Nebraska Press: 207-283.
Ekman, P. (1999). Basic Emotions. In T. Dalgleish and T. Power (Eds.) The Handbook of Cognition and Emotion. Pp. 45–60. Sussex, U.K.: John Wiley & Sons, Ltd.
Godden, D. R. & Baddeley, A. D. (1975). Context-Dependent Memory in Two Natural Environments: On Land and Underwater. British Journal of Psychology, 66: 325–331.
Godden, D. and Baddeley, A. (1980), When does context influence recognition memory?. British Journal of Psychology, 71: 99–104.
Goodwin, D.W., Powell, B., Bremer, D., Hoine, H. & Stern, J., “Alcohol and Recall: State-Dependent Effects in Man”, Science, (1969), Vol.163, p.1358.Immordino-Yang & Damasio, 2007
Isen, A. M. & Shalker, T. E. (1982). The effect of feeling state on evaluation of positive, neutral, and negative stimuli: When you “accentuate the positive,” do you “eliminate the negative”? Social Psychology Quarterly, 45 (1), pp. 58–63.
Loftus, E.F., Banaji, M.R., Schooler, J.W., & Foster, R.A. (1987). Who remembers what? Gender differences in memory. Michigan Quarterly Review, 26, 64-85.
Marian, V. & Kaushanskaya, M., “Language context guides memory content”, Psychonomic Bulletin and Review, 14(5), (2007), pp.925–933.
Martini, F. & Bartholomew, E. (2003). Essentials of Anatomy & Physiology. San Francisco: Benjamin Cummings. p. 267.
Nock, M. K., Holmber, E. B., Photos, V. I., & Michel, B. D. (2007). Self-Injurious Thoughts and Behaviors Interview: Development, reliability, and validity in an adolescent sample. Psychological Assessment, 19(3), 309-31.
Overton, D.A., “Dissociated learning in Drug States (State-Dependent Learning)”, pp.918-930, in Efron, D.H., Cole, J.O., Levine, J. & Wittenborn, R. (eds.), Psychopharmacology: A Review of Progress 1957-1967, U.S. Government Printing Office, (Washington), 1968.
Posner, M. I. (1978). Chronometric explorations of mind. Hillsdale, NJ: Erlbaum, 1978
Posner, M.I. & Rothbart, M.K. (2000). Developing mechanisms of self-regulation. Development and Psychopathology, 12 , pp 427-441
Snow, W.G., Tierney, M.C., Zorzitto, M.L., Fisher, R.H., & Reid, D.W. (1988). One year test-retest reliability of selected neuropsychological tests in older adults. Paper presented to the International Neuropsychological Society, New Orleans.
Strongman, K. & Russell, P. (1986). Salience of Emotion and Recall. Bulletin of the Psychonomic Society 24(1): 25-27.
Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1997). Cognitive psychology and emotional disorders (2nd ed.). Chichester: Wiley.
The Blog of Frank Kotsianas 