ABSTRACT
The attentional blink (AB) is the impaired ability to detect a second target (T2) when it follows shortly after the first (T1) among distractors in a rapid serial visual presentation (RSVP). Given questions about the automaticity of age differences in emotion processing, the current study examined whether emotion cues differentially impact the AB elicited in older and younger adults. Twenty-two younger (18–22 years) and 22 older adult participants (62–78 years) reported on the emotional content of target face stimulus pairs embedded in a RSVP of scrambled-face distractor images. Target pairs included photo-realistic faces of angry, happy, and neutral expressions. The order of emotional and neutral stimuli as T1 or T2 and the degree of temporal separation within the RSVP systematically varied. Target detection accuracy was used to operationalise the AB. Although older adults displayed a larger AB than younger adults, no age differences emerged in the impact of emotion on the AB. Angry T1 faces increased the AB of both age groups. Neither emotional T2 attenuated the AB. Negative facial expressions held the attention of younger and older adults in a comparable manner, exacerbating the AB and supporting a negativity bias instead of a positivity effect in older adults.
Acknowledgement
Research conducted by Allison M. Sklenar and Andrew Mienaltowski at the Department of Psychological Sciences, Western Kentucky University in conjunction with the first author’s master’s thesis. Data were presented in part as a poster at the annual meeting of the Association for Psychological Science in 2016. Allison M. Sklenar is now at the Department of Psychology, University of Illinois at Chicago.
Disclosure statement
No potential conflict of interest was reported by the authors.
ORCID
Allison M. Sklenar http://orcid.org/0000-0001-8255-4709
Andrew Mienaltowski http://orcid.org/0000-0003-0762-8311
Notes
1 Because the current study included younger and older adults, the overall sample size is twice that of Mather and Carstensen (Citation2005; see also Anderson, Citation2005; Bach et al., Citation2014; Chun & Potter, Citation1995; Hommel & Akyürek, Citation2005; Ogawa & Suzuki, Citation2004). Power analyses using G*Power software (Faul, Erdfelder, Lang, & Buchner, Citation2007) indicated that this sample size allows for a power of .80 or greater to detect a younger adult advantage over older adults at a large effect size ( = .11 and up), to detect differences between lags at a medium effect size (
= .05 and up), to detect emotion type differences at a medium-to-large effect size (
= .11 and up), to detect age group × emotion interactions at a medium-to-large effect size (
= .11 and up), age group × lag interactions at a medium effect size (
= .05 and up), and emotion type × lag interactions at a small-to-medium effect size (
= .03 and up).
2 NimStim Models- 1, 3, 5–14, 16, 18–26, 28, 29, 31–40, 43. Karolinska Models- Set A Males: 2, 9, 14, 21, 22, 29; Set B Males: 1, 5, 6, 10, 16, 21, 22, 24, 25, 28, 30, 31; Set A Females: 1, 8, 9, 14, 17, 19–23, 26, 32, 33. Set B Females: 7, 9, 16, 20, 24, 25, 35.
3 Using the same factors, a Bayesian mixed-model ANOVA performed in JASP demonstrated that the data for the emotional-neutral trials are explained significantly better than a null model by a model that includes main effects of age group (B = 3.9e+6), trial type (B = 8.5e+13), and lag (B = ∞), as well as age group × lag (B = 2.1e+5) and trial type × lag (B = 273) interactions. Each Bayes factor for the aforementioned terms provides extreme evidence supporting the inclusion of each term. In the model comparison within the analysis, the excluded terms did not improve the explanation of the data over prior models by a factor greater than 10 (Wagenmakers et al., Citation2018), and had Bayesian factors below 0.33 (Dienes, Citation2014), trial type × age group (B = 0.058) and age group × trial type × lag (B = 0.005). The Bayes factors for the excluded terms provide strong evidence and extreme evidence, respectively, for null effects of these excluded terms, and do not suggest that the data are insensitive to age group × trial type interactions.
4 As for the emotional-neutral trials, a Bayesian mixed-model ANOVA performed in JASP demonstrated that the data for the neutral-emotional trials are explained significantly better by a model that includes main effects of age group (B = 8.1e+10), trial type (B = 3.5), and lag (B = ∞), as well as an age group × lag (B = 4.3e+9) interaction than by a null model. Note that the Bayes factors suggest that there is extreme evidence to support each of the aforementioned terms except the main effect of Trial Type, which only garnered moderate evidence. The Bayesian ANOVA did not replicate a significant Trial Type × Lag (B = 0.016) interaction, consistent with how the data have been interpreted here. Each excluded model term failed to improve the explanation of the data over prior models by a factor greater than 10 and had Bayesian factors that provided strong to extreme evidence for null effects (Wagenmakers et al., Citation2018): trial type × lag (B = 0.016), trial type × age group (B = 0.082), and age group × trial type × lag (B = 8.1e−5). The Bayes factors for these excluded model terms support null outcomes rather than insensitivity to observing significant terms.
5 Number question accuracy may give some insight into memory for perceiving faces without the detailed memory for the specific emotions expressed by those faces. In other words, accuracy for the number of faces present in each trial independent of memory for the emotions might reflect whether or not faces were noticed, which may very well depend upon emotion, even if not distinctly recollected. For emotional-neutral trials, an ANOVA on number question accuracy revealed main effects of age group, F(1, 42) = 8.11, p = .007, = .162, trial type, F(2, 84) = 7.41, p = .001,
= .150, and lag, F(7, 294) = 47.34, p < .001,
= .530, and interactions for lag × age group, F(7, 294) = 12.11, p < .001,
= .224, and trial type × lag, F(14, 588) = 3.40, p < .001,
= .075. The lag × age group and trial type × lag interactions mirrored those observed for RSVP accuracy. In addition, lag-1 sparing was observed for number question accuracy for younger adults regardless of trial type (i.e., emotion). Number accuracy was also examined for neutral-emotional trials. An ANOVA revealed main effects of trial type, F(2, 84) = 11.43, p < .001,
= .214, and age group, F(1, 42) = 9.63, p < .001,
= .186, and interactions for trial type × lag, F(14, 588) = 1.818, p = .033,
= .041, lag × age group, F(7, 294) = 14.72, p < .001,
= .259, and trial type × lag, F(14, 588) = 1.82, p = .033,
= .041. The trial type × lag interaction emerged because there was greater improvement from Lag 1 to Lag 2 for neutral-angry (MLag1 = 58.4, SELag1 = 3.3; MLag2 = 75.2, SELag2 = 3.7) and neutral-happy (MLag1 = 57.7, SELag1 = 3.8; MLag2 = 72.2, SELag2 = 4.2) than for neutral-neutral trials (MLag1 = 57.5, SELag1 = 3.4; MLag2 = 64.1, SELag2 = 4.4), which supports an emotional enhancement effect. The lag × age group interaction mirrored that observed in the RSVP accuracy already reported, again with the addition of lag-1 sparing for number question accuracy for younger adults. Lastly, the age group × trial type interaction emerged because younger adults displayed greater overall accuracy on neutral-happy trials (M = 88.6, SE = 2.3) than on neutral-neutral trials (M = 85.5, SE = 2.4), whereas older adults displayed an emotion enhancement effect, or greater accuracy on both neutral-angry (M = 76.0, SE = 3.6) and neutral-happy trials (M = 73.9, SE = 4.0) than neutral-neutral trials (M = 70.2, SE = 4.2).