Changing interpretations of emotional expressions in working memory with aging

Abstract

[Correction Notice: An Erratum for this article was reported online in Emotion on Jun 24 2019 (see record 2019-34942-001). In the article, the plots for Figure 3a shifted incorrectly to the right. The error bars should be centered on 10, 30, 50, 70, and 90. The corrected figure is present in the erratum.] Working memory (WM) shows significant decline with age. It is interesting to note that some research has suggested age-related impairments can be reduced in tasks that involve emotion-laden stimuli. However, only a few studies have explored how WM for emotional material changes in aging. Here we developed a novel experimental task to compare and contrast how emotional material is represented in older versus younger adults. The task enabled us to separate overall WM accuracy from emotional biases in the content of affective representations in WM. We found that, in addition to overall decline in WM performance, older adults showed a systematic positivity bias in representing information in WM relative to younger adults (positivity effect). They remembered fearful faces as being less fearful than younger adults and interpreted ambiguous facial expressions more positively. The findings show that aging brings a type of positivity bias when picking up affective information for guiding future behavior. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Figure 1

WM task schematic and WM error results. In the WM task (a), participants encoded a facial expression into WM, and maintained it over a delay of 3,000 ms. A test face with the same facial identity but a neutral facial expression (0% intensity) appeared, and participants changed the face to match the expression intensity in memory using a trackball mouse. Target faces were fearful or happy faces from 0% to 100% in emotional intensity. Emotion type was intermixed within blocks. Bar plots in (b) show WM error for fearful (red, left) and happy faces (blue, right) in the young and old participant groups. Error bars represent SEM. *** p < .001, * p < .05. Faces presented are part of the NimStim stimulus set, for which use for publication is permitted.

Figure 2

Emotional bias in WM. Responses are plotted as a function of the target-face emotion type and emotional intensity in (a), with negative values representing intensity values of fearful faces and positive values representing intensity values of happy faces. Responses were binned into five equal bins for fearful faces (from −100 to −20% in 20% steps, with the 20% bin including −20 to −1%) and five bins for happy faces (from 20% to 100% in 20% steps) and a 0% bin with only neutral faces for visualization. Perfect performance corresponds to responses on the diagonal (dotted line). On the right side of zero (y axis), responses above the line mean that faces were reported to be happier than happy-face targets, whereas responses below the line mean that faces were reported to be less happy than targets. On the left side of zero, responses below the line mean that faces were reported to be more fearful than fearful-face targets, whereas responses above the line mean that faces were reported as less than fearful-face targets. The bias is shown in (b), computed by taking the mean of each participant’s raw psychometric curve—note that (a) is binned for visualization. Bias for each of the emotion types is plotted in (c). Responses for fearful faces were flipped to have a positive sign, and trials with neutral faces were excluded. Mean response was computed for each emotion type (from 1% to 100%) and normalized by subtracting 50 (see Data Analysis for details). *** p < .001.

Figure 3

Older adults interpreted fearful faces with low emotional intensities as happy more than younger adults. Proportion of trials correctly judged as fearful in the WM task is plotted for each emotional intensity bin from 1% to 80% in 20% steps for younger and older participants in the top panel in (a). An illustration showing how low-to-medium fearful faces are sometimes judged as happy faces in the bottom panel of (a). Proportion of trials correctly judged as happy in the WM task are plotted for each emotional intensity bin in the top panel of (b) for younger and older participants, with an illustration in the bottom panel showing how low-to-medium happy faces are sometimes judged as fearful faces. Faces presented are part of the NimStim stimulus set, for which use for publication is permitted.