Contextual Changes Influence Attention Flexibility Towards New Goals

Abstract

The ability to flexibly process affective information underlies resilient adaptation to changing situations. However, the impact of social contextual changes on affective flexibility has not been examined yet, although this may be crucial to understand how attention operates when changing situations require allocation of different emotional goals. In a novel eye-tracking task, participants had to deploy the goals to attend to positive or negative facial expressions based on contextual cues. We found that dysphorics, compared to non-dysphorics, were faster in switching to negative goals, but slower in switching to positive goals, when the context remained constant. However, when the context changed, dysphorics showed an even faster switch to negative goals, while non-dysphorics became more adept at switching towards positive goals. These results suggest that contextual changes exacerbate the negativity and positivity biases exhibited by dysphoric and non-dysphoric individuals, respectively. The study provides evidence for the key role played by contextual changes in guiding attention allocation and thereby emotion regulation processes. These findings provide preliminary evidence for the role of contextual inflexibility in dysphoria. Results need to be replicated in a clinical sample to further clarify the role of inflexibility over the course of depression.

Appendices

Appendix I

A description of the 4 social contexts developed for this study. Participants were presented with the text of each social context which briefly explained what took place in the context, along with the two goals that could be activated in each context. Along with the text, participants were presented with the 2 goal-activating objects, square and circle, which were associated with different goals in different contexts. Lastly, participants were also presented the context picture which represented the textual context in a pictorial form.

Appendix II

Here, we present the 4-way interaction decomposition below:

Source	Test-value	p-value	Effect size
2 (context switch) × 2 (goal switch) × 2 (goal type) × 2 (group)	F(1, 49) = 4.59	p < 0.05*	$\eta_{p}^{2} = .09$
2 (context switch) × 2 (goal switch) × 2 (group) (Goal type level: look to positive)	F(1, 49) = 2.04	p = 0.16	$\eta_{p}^{2} = .04$
2 (context switch) × 2 (goal switch) × 2 (group) (Goal type level: look to negative)	F(1, 49) = 0.65	p = 0.42	$\eta_{p}^{2} = .01$
2 (context switch) × 2 (goal type) × 2 (group) (goal switch level: repeat goal)	F(1, 49) = 2.46	p = 0.12	$\eta_{p}^{2} = .05$
2 (context switch) × 2 (goal type) × 2 (group) (Goal switch level: switch goal)	F(1, 49) = 19.24	p < 0.001**	$\eta_{p}^{2} = .28$
2 (context switch) × 2 (group) (Goal type level: look to positive)	F(1, 49) = 18.36	p < 0.001**	$\eta_{p}^{2} = .27$
2 (context switch) × 2 (group) (goal type level: look to negative)	F(1, 49) = 2.84	p = 0.056	$\eta_{p}^{2} = .07$
T-test for variable context switch (level: switch context)	t(49) = 0.5.29	p < 0.001**	d = 1.47
T-test for variable context switch (level: repeat context)	t(49) = 0.32	p = 0.752	d = 0.09

Here, we present the 3-way interaction decomposition for the high dysphoric group:

Source	Test-value	p-value	Effect size
2 (context switch) × 2 (goal switch) × 2 (goal type)	F(1, 23) = 14.42	p = 0.001**	$\eta_{p}^{2} = .39$
2 (context switch) × 2 (goal switch) (Goal type level: look to positive)	F(1, 23) = 7.20	p = 0.013*	$\eta_{p}^{2} = .24$
2 (context switch) × 2 (goal switch) (Goal type level: look to negative)	F(1, 23) = 40.60	p < 0.001**	$\eta_{p}^{2} = .64$
2 (context switch) × 2 (goal type) (Goal switch level: repeat goal)	F(1, 23) = 11.97	p = 0.002*	$\eta_{p}^{2} = .34$
2 (context switch) × 2 (goal type) (Goal switch level: switch goal)	F(1, 23) = 4.54	p < 0.04*	$\eta_{p}^{2} = .17$

Here, we present the 3-way interaction decomposition for the low dysphoric group:

Source	Test-value	p-value	Effect size
2 (context switch) × 2 (goal switch) × 2 (goal type)	F(1, 26) = 9.49	p = 0.005*	$\eta_{p}^{2} = .27$
2 (context switch) × 2 (goal switch) (Goal type level: look to positive)	F(1, 26) = 24.88	p < 0.001**	$\eta_{p}^{2} = .49$
2 (context switch) × 2 (goal switch) (Goal type level: look to negative)	F(1, 26) = 52.90	p < 0.001**	$\eta_{p}^{2} = .67$
2 (context switch) × 2 (goal type) (Goal switch level: repeat goal)	F(1, 26) = 140.65	p = 0.002*	$\eta_{p}^{2} = .84$
2 (context switch) × 2 (goal type) (Goal switch level: switch goal)	F(1, 26) = 21.29	p < 0.001**	$\eta_{p}^{2} = .45$

Bayesian Analyses

We conducted analyses using JASP (version 0.9.2.0; 2019). We used BF10, which is the probability of the alternative hypothesis relative to the null hypothesis, and we employed default priors (Cauchy scale: 0.0707) in our analyses. We interpreted the Bayes factors as follows: BF10 < 0.01 = decisive evidence for H0; BF10 = 0.01–0.033 = very strong evidence for H0; BF10 = 0.033–0.1 = strong evidence for H0; BF10 = 0.1–0.33 = substantial evidence for H0; BF10 = 0.3–1 = anecdotal evidence for H0; BF10 = 1 = no evidence for either hypothesis; BF10 = 1–3 = anecdotal evidence for H1; BF10 = 3–10 = substantial evidence for H1; BF10 = 10–30 = strong evidence for H1; BF10 = 30–100 = very strong evidence for H1; BF10 > 100 = decisive evidence for H1.

First Hypothesis

When the context was constant (i.e., repeated from the previous trial), using Bayesian paired t-test comparison we found strong evidence (BF = 53.81) suggesting that high dysphoric individuals had faster first fixations when switching from a positive to a negative goal in comparison to switching from a negative to a positive goal. Similarly, when the context remained constant, we found substantial evidence (BF = 7.30) indicating that high dysphoric individuals had faster first fixations when repeating attention to a negative goal in comparison to a positive goal. These findings are consistent with the frequentist analysis.

Second Hypothesis

Using Bayesian paired t-test comparison, we found substantial evidence for the null (BF = 0.24) indicating that for low dysphorics there were no significant differences in first fixations when repeating attention to negative or positive goals, when the context was constant. This finding is consistent with the frequentist analysis. However, we found anecdotal evidence in favor of the null (BF = 0.49) indicating that we could not claim with certainty that there were no significant differences for low dysphorics in switching attention to positive or negative goals when the context was constant. This finding deviates from the result from the frequentist analysis, wherein we found a statistically significant result for no differences in switching towards positive or negative goals when context is repeating. This discrepancy suggests that we cannot ultimately conclude with certainty, and replication of the study is needed in order to clarify this finding.

Third Hypothesis

Repeating Goals

In line with our third hypothesis that contextual changes would create difficulties for all individuals, using Bayesian paired t-test comparisons we found very strong to decisive evidence indicating that high dysphorics were slower when repeating attention to positive (BF = 50.70) and negative (BF > 100) goals. Further, we found decisive evidence (BF > 100) indicating that low dysphoric individuals were slower when repeating attention to negative goals. These findings are consistent with the frequentist analysis. However, for the comparison between context repeat vs. context switch when repeating attention to positive we found anecdotal evidence (BF = 2.81) for low dysphorics. This suggests that we cannot claim that low dysphorics were significantly slower in repeating attention to positive goals when the context switched versus repeating. This finding deviates somewhat from the frequentist analysis, wherein we found that low dysphorics were significantly slower to repeat attention to positive goals when context switched vs. when context repeated. Once again, this discrepancy suggests that we cannot ultimately conclude with certainty, and replication of the study is needed in order to clarify this finding.

Switching Goals

Bayesian paired t-test comparisons provided substantial evidence (BF = 3.04) indicating that high dysphoric individuals had faster fixations when switching attention from a positive to a negative goal when the context was switching as compared to when the context was repeating. However, when switching from negative to positive goals, we found substantial evidence for the null (BF = 0.15), indicating that there were no significant differences for high dysphorics in fixation latencies when the context was switching vs. when the context was repeating. In contrast, we found decisive evidence (BF > 100) indicating that low dysphoric individuals had faster first fixations when switching attention from a negative to a positive goal when the context was switching as compared to when the context was repeating. However, when switching from a positive to a negative goal, we found substantial evidence for the null (BF = 0.28), suggesting a consistent lack of differences in fixation times when the context was switching vs. repeating. These results are in line with our findings from the frequentist analysis, as reported in the manuscript.

Fourth Hypothesis

For conditions where the context switched (i.e., changed from the previous trial), pairwise comparisons found decisive evidence indicating that high, compared to low dysphorics, found it more difficult to both repeat attention (BF > 100) and switch attention (BF > 100) towards positive goals. Meanwhile, we only found anecdotal evidence (BF = 2.67) indicating that high dysphorics had more difficulties compared to low dysphorics in repeating attention to negative goals when context was switching. Finally, we found substantial evidence for the null (BF = 0.01) when switching towards negative goals. This indicates that high dysphorics, compared to low dysphorics, were faster in switching attention from positive towards negative goals when the context was switching. Again, these results are consistent with findings from the frequentist analysis.

Appendix III

This table provides the correlation between the scores on the Anhedonic Depression subscale of Mood and Anxiety Symptoms Questionnaire-D30 (MASQ-D30) and the 8 attention switching indices. In the second column correlations are provided for attention switching indices and the overall scores on MASQ-D30. **Correlation is significant at the 0.01 level (2-tailed). *Correlation is significant at the 0.05 level (2-tailed).

Attention switching index	Anhedonic Depression subscale (p-value)	Total MASQ-D30 (p-value)
Context repeat–goal repeat (look to negative)	0.14 (p = 0.32)	0.07 (p = 0.62)
Context repeat–goal repeat (look to positive)	0.31* (p = 0.029)	0.27 (p = 0.056)
Context repeat–goal switch (look to negative)	− 0.54** (p < 0.001)	− 0.38** (p = 0.006)
Context repeat–goal switch (look to positive)	0.17 (p = 0.25)	0.23 (p = 0.11)
Context switch–goal repeat (look to negative)	0.19 (p = 0.17)	0.18 (p = 0.19)
Context switch–goal repeat (look to positive)	0.68** (p < 0.001)	0.55** (p < 0.001)
Context switch–goal switch (look to negative)	− 0.60** (p < 0.001)	− 0.48** (p < 0.001)
Context switch–goal switch (look to positive)	0.35* (p = 0.012)	0.29* (p = 0.035)

Appendix IV

During the attention flexibility task, participants were introduced to the 4 contexts, their associated goals, and the corresponding context picture and the goal objects. Participants memorized this information, which was then presented in the form of a reaction-time task using eye-tracking. Each trial for the task began with a fixation cross for 500 ms. This was replaced by a picture of a context with a goal object on top of it. After fixation on the goal object for 100 ms, participants saw the sentence “Press spacebar for faces”. At this point participants could take as long as they needed to recall the correct response (attend to positive or attend to negative), based upon the context picture and the goal object combination presented on the screen. Once the participants were ready, they pressed the spacebar and saw a fixation cross for 500 ms, followed by 8 emotional faces around the cross.

Data Preparation

In order to assess how quickly participants were able to mental-shift between different contextual goals, we measured the time it took participants to press the spacebar. We obtained the total presentation time of context and goal cue for each trial, measured from the moment the context picture appeared on the screen with the goal object on top of it until the moment participants pressed the spacebar to view the faces. Total cue presentation times were obtained for each of the 128 trials. From the total cue presentation times we were able to calculate the set-shifting variables. We calculated 8 set-shifting variables: repeat context–repeat goal (look to positive), repeat context–repeat goal (look to negative), repeat context–switch goal (look to positive), repeat context–switch goal (look to negative), switch context–repeat goal (look to positive), switch context–repeat goal (look to negative), switch context–switch goal (look to positive), and switch context–switch goal (look to negative). This index allowed us to ascertain how quickly participants were able to mentally set-shift between the different context.

A 2 (context switch) × 2 (goal switch) × 2 (goal) × 2 (group) mixed analysis of variance (ANOVA) was conducted with the ‘Total cue presentation time’ variable as dependent variable. Group (high vs. low dysphorics) was the between-subjects factor. Context switch (repeat vs. switch context), goal switch (repeat vs. switch goal), and goal (look to positive vs. look to negative) were the within-subjects factors. We found significant main effects for context switch, F(1, 49) = 100.89, p < 0.001, $\eta_{p}^{2} = .67$, goal switch, F(1, 49) = 43.95, p < 0.001, $\eta_{p}^{2} = .47$, and goal, F(1,49) = 34.75, p < 0.001, $\eta_{p}^{2} = .42$. These main effects were qualified by a significant three-way context switch × goal switch × group interaction, F(1, 49) = 6.62, p = 0.013, $\eta_{p}^{2} = .12$.

In order to decompose the significant 3-way interaction above, we conducted four separate 2 (context switch) × 2 (group) mixed ANOVAs for each level of goal switch. We found 4 significant interactions. First, we found significant context switch × group interactions for the ‘goal repeat’ level when the goal was positive, F(1,49) = 23.12, p < 0.001, $\eta_{p}^{2} = .32$, and when the goal was negative, F(1,49) = 40.14, p < 0.001, $\eta_{p}^{2} = .45$. Next, we found significant context switch × group interactions for the ‘goal switch’ level when the goal was positive, F(1,49) = 33.82, p < 0.001, $\eta_{p}^{2} = .41$, and when the goal was negative, F(1,49) = 56.79, p < 0.001, $\eta_{p}^{2} = .54$.

Significant Bonferroni-corrected pairwise comparisons revealed that high dysphorics had longer presentation times (i.e. they took longer to press the spacebar) when the context switched, compared to when the context was repeating, both while repeating positive goals (p < 0.001) and while switching towards positive goals (p < 0.001). Meanwhile within-group comparisons showed that low dysphorics did not have significantly longer presentation times (i.e. they were faster in pressing the spacebar) when context switched, compared to when the context was repeating, both while repeating positive goals (p = 0.121) and while switching towards positive goals (p = 0.128). Further, significant within-group comparisons revealed that high dysphorics also had longer presentation times when the context switched, as opposed to when it repeated, both while repeating negative goals (p < 0.001) and when switching towards negative goals (p < 0.001). Low dysphorics, on the other hand, had significantly longer presentation times when the context switched, compared to when the context repeated, when repeating negative goals (p = 0.039). However, low dysphorics did not have significantly longer presentation times when the context switched, while switching attention towards negative goals (p = 0.230). Lastly, significant pairwise comparison revealed that high compared to low dysphoric individuals had longer presentation times when the context switched, both while repeating and switching to both positive and negative goals (all ps < 0.001; see Fig. 5 below).

Fig. 5

Reaction times (i.e., time to press spacebar) are presented for both high and low dysphorics for the four different conditions, combinations of context switch (repeat vs. switch context) and goal switch (repeat vs. switch goal). Times are presented in seconds (along the y-axis). High = high dysphoric, low = low dysphoric