Article Figures & Data
Figures
- Figure 1.
Experimental design. A, On each trial, during the initial choice stage, participants viewed images of chairs and planes displayed side-by-side on the screen for 2 s and were asked to select one of those images. A yellow-colored square highlighted the selected choice. After a jittered ISI, an emotional image (positive or neutral) was shown as feedback during the subsequent outcome stage (2 s). The valence of the emotional image signaled monetary loss or no-loss outcome. In the example trial shown, if participants saw a positive image, it indicated that they lost money on that trial, while neutral indicated no loss. This emotion–loss mapping was reversed in the other phase. Each trial ended with a blank jittered ITI period. B, In one phase (Phase 1), positive stimulus indicated a loss outcome, and neutral stimulus indicated a no-loss outcome. This valence–outcome mapping was reversed in the second phase (Phase 2). Each participant completed both phases of the experiment, and the order of phases was counterbalanced across participants. ISI, interstimulus interval; ITI, intertrial interval.
- Figure 2.
Main effects of emotion and loss. Clusters that exhibited the main effects of (A) Emotion and (B) Loss outcomes at an uncorrected p = 0.001 and 37-voxel cluster extent (cluster-level α of 0.05). Because cluster-extent–based thresholding was used for multiple-comparison correction, voxels are displayed using a binary threshold. For further rationale about using binary maps in the context of cluster-based thresholding, see Woo et al., 2014.
- Figure 3.
Conjunction of emotion and loss effects. Clusters that exhibited conjunction of the main effects of Emotion and Loss factors in (A) vmPFC, (B) bilateral caudate, (C) bilateral amygdala, and (D) bilateral lateral occipital cortex. For each cluster, the bar plot shows the average group-level estimates of the four outcome conditions of interest. It should be noted that the primary purpose of these average bar plots is to illustrate the overlapping pattern of the main effects and not to draw any statistical inference. Hence, error bars were not plotted on top of the average estimates to avoid issues of circularity (Kriegeskorte et al., 2009).
- Figure 4.
Relationship between emotion and loss effects. Pearson's correlation between Emotion and Loss main effects across participants in clusters that exhibited conjunction of the main effects in (A) vmPFC, (B) bilateral caudate, and (C) bilateral amygdala. Emotion main effect index: (loss + no loss)Positive – (loss + no loss)Neutral; Loss main effect index: (positive + neutral)Loss – (positive + neutral)No loss.
- Figure 5.
Subjective hedonic ratings of experienced outcomes. The bar plot shows the average self-reported pleasantness ratings for the four outcome conditions on a scale of 1–9 (1 being unpleasant and 9 being pleasant). Each dot indicates the individual participant ratings.
- Figure 6.
Estimated learning rates from the RL model. A, The bar plot shows the average estimated learning rates for the chair and plane stimulus for two phases separately. In phase 1, positive stimulus signaled a loss outcome and neutral stimulus signaled a no-loss outcome. In phase 2, neutral stimulus signaled a loss outcome and positive stimulus signaled a no-loss outcome. B, The bar plot shows the average estimated learning rates for the deemed high inherent valued and low inherent valued stimulus for two phases separately. Each dot indicates the estimated learning rate of each participant.
Tables
- Table 1.
Mean self-reported valence and arousal ratings (with standard deviation in parentheses) of the two sets of 36 scene stimuli each that were employed in this study
Valence ratings Descriptive values Two-way ANOVA results Set1 Set2 Effect F value p value Emotion Mean SD Mean SD Main effect: emotion 41.391 <0.001 Positive 7.19 0.22 7.16 0.19 Main effect: set 1.862 0.202 Neutral 5.22 0.29 5.16 0.25 Interaction effect: emotion*set 0.094 0.766 Arousal ratings Descriptive values Two-way ANOVA results Set1 Set2 Effect F value p value Emotion Mean SD Mean SD Main effect: emotion 134.683 <0.001 Positive 6.28 0.98 6.16 1.18 Main effect: set 6.537 0.029 Neutral 1.88 0.36 1.75 0.36 Interaction effect: emotion*set 0.002 0.969 A 2 emotion (positive, neutral) × 2 set (Set1, Set2) two-way ANOVA was performed on the valence and arousal ratings of the scene stimuli. A main effect of emotion was observed in both the valence and arousal ratings with the positive scenes being rated higher in both valence and arousal than the neutral scenes. No emotion × set interaction was observed for both valence and arousal ratings.
Positive > neutral
ES = [(loss + no loss)Positive – (loss + no loss)Neutral]
Cluster x y z k F(1,34) ES Medial frontal cortex, cingulate cortex, precuneus, fusiform gyrus, cerebellum, thalamus, brainstem, midbrain 56 −72 −4 42,267 50.69 1.36 R. precentral gyrus 50 6 36 1,622 47.14 0.46 L. precentral gyrus −50 0 34 1,119 23.02 0.26 R. inferior frontal gyrus 54 40 10 231 26.93 0.34 R. middle temporal gyrus 64 0 −22 177 14.03 0.14 L. anterior insula −40 26 4 146 60.66 0.14 L. inferior frontal gyrus −52 40 10 146 20.17 0.19 R. superior frontal gyrus 2 10 62 124 14.69 0.11 L. perirhinal/rhinal sulcus −38 −4 −34 111 25.33 0.16 L. middle insula −36 −6 18 61 35.66 0.11 R. globus pallidus 22 −2 8 53 14.45 0.06 R. middle insula 38 −2 14 38 27.44 0.12 Neutral > positive
ES, [(loss + no loss)Neutral – (loss + no loss)Positive]
Brain regions x y z k F(1,34) ES R. parahippocampal gyrus 28 −46 −10 537 168.18 0.30 L. parahippocampal gyrus −28 −60 −10 479 70.27 0.25 R. angular gyrus 52 −60 50 433 17.17 0.18 R. medial frontal gyrus, inferior frontal sulcus 42 26 40 336 24.26 0.16 L. anterior transverse temporal gyrus −56 −16 10 271 18.33 0.14 R. superior temporal gyrus 68 −10 −6 196 14.59 0.18 R. superior frontal gyrus 4 28 44 155 37.62 0.11 R. transverse temporal gyrus 50 −28 14 110 19.21 0.14 R. superior/middle temporal gyrus −46 −32 8 64 20.73 0.11 Calcarine sulcus/striate cortex 0 −84 22 37 13.24 0.11 The following clusters survived multiple-comparison correction in univariate whole-brain voxel-wise analysis at the outcome phase [peak MNI coordinates, cluster size (k), F(1,34) values, and unstandardized effect size in terms of % signal change (ES)].
No loss > loss
ES = [(positive + neutral)No loss – (positive + neutral)Loss]
Cluster x y z k F(1,34) ES Bilateral striatum −6 0 4 3,936 15.23 0.35 Medial frontal cortex (vmPFC) 0 62 4 2,385 31.57 0.24 R. cerebellum (lateral zone) 48 −72 −42 2,158 40.87 0.29 R. middle/inferior temporal gyrus 60 −56 −16 1,171 14.47 0.20 R. inferior occipital gyrus 26 −98 −12 863 19.52 0.29 L. middle temporal gyrus −64 −54 −8 791 13.58 0.19 L. inferior occipital gyrus −6 −98 −18 763 13.31 0.32 L. middle/inferior frontal gyrus −50 40 18 693 15.05 0.15 L. cerebellum (lateral zone) −42 −74 −48 658 16.67 0.21 R. posterior cingulate cortex/precuneus 0 −46 38 459 14.06 0.13 R. anterior orbital gyrus 40 40 −20 394 16.50 0.19 Cerebellum (vermis) −2 −42 −36 354 14.16 0.17 L. precentral gyrus, inferior precentral sulcus −54 8 42 223 15.03 0.19 R. superior frontal gyrus (lateral part), accessory superior frontal sulcus 26 30 52 204 22.08 0.09 Midcingulate cortex 0 −8 28 155 15.53 0.12 L. cerebellum (posterior lobe) −20 −86 −46 144 13.31 0.22 L. lateral posterior Orbital gyrus, sulcus −20 36 −22 133 13.33 0.14 L. inferior frontal gyrus −60 12 22 129 16.33 0.12 L. superior frontal gyrus, paracingulate 0 30 42 120 14.05 0.13 L. angular gyrus −50 −70 42 106 13.29 0.16 L. cerebellum (anterior lobe) −10 −84 −24 105 14.40 0.13 Cerebellum (posterior lobe, vermis) 4 −58 −22 100 17.58 0.09 L. superior temporal gyrus −68 −28 14 90 14.48 0.18 L. paracentral lobule −2 −24 58 83 14.28 0.10 L. superior frontal gyrus, accessory superior frontal sulcus −22 32 56 80 14.66 0.12 L. cerebellum (anterior lobule) −10 −36 −16 75 13.13 0.11 R. cerebellum (anterior lobule) 18 −28 −28 69 13.48 0.11 Cerebellum (central lobule) −4 −44 −8 61 20.10 0.09 L. striate cortex −2 −102 12 60 13.41 0.19 L. supramarginal gyrus/superior parietal lobule −34 −44 38 57 18.34 0.08 L. middle temporal gyrus, superior temporal sulcus −58 −38 0 46 13.77 0.08 R. paracentral lobule 4 −28 58 40 14.26 0.10 L. precentral gyrus −10 −30 66 39 27.56 0.08 L. lateral caudate −18 −6 26 38 21.48 0.18 Loss > no loss
ES = [(positive + neutral)Loss–(positive + neutral)No Loss]
No clusters found. The following clusters survived multiple-comparison correction in univariate voxel-wise analysis at the outcome phase [peak MNI coordinates, cluster size (k), F(1,34) values, and unstandardized effect size in terms of % signal change (ES)].
Cluster x y z k BF01 vmPFC 2 26 −28 1,895 4.39 R. lateral occipital cortex 30 −92 −14 764 1.23 R. cerebellum (lateral zone) 18 −76 −46 562 3.05 R. amygdala 32 −4 −22 486 3.87 L. lateral occipital cortex −32 −96 −14 451 1.70 L. amygdala −26 −4 −20 241 3.51 L. posterior Cingulate cortex −8 −54 20 212 2.21 R. lateral Orbitofrontal cortex 30 36 −20 169 3.90 Midcingulate cortex −2 6 26 141 2.32 R. inferior temporal gyrus/fusiform gyrus 46 −50 −20 123 2.83 Cerebellum (posterior lateral fissure) −10 −46 −48 120 1.66 L. lateral Orbitofrontal cortex −20 28 −18 120 2.36 R. caudate/ventral Striatum 6 8 −10 113 1.53 Inferior thalamic peduncle 6 −4 2 94 1.28 L. intermediate Frontal sulcus −52 36 4 83 1.28 L. caudate/ventral Striatum −4 4 −8 80 1.08 L. inferior frontal gyrus −52 6 20 69 3.94 Cerebellum (vermis) 0 −60 −32 60 4.09 L. cerebellum (lateral zone) −30 −74 −42 53 2.44 L. cerebellum (intermediate zone) −6 −80 −49 52 2.11 R. posterior cingulate cortex 8 −52 18 49 2.04 L. inferior precentral sulcus −42 4 36 49 1.33 R. middle temporal gyrus 62 −2 −24 40 4.30 L. supramarginal gyrus −34 −44 34 39 2.32 The following clusters survived the multiple-comparison correction in univariate voxel-wise conjunction analysis at the outcome phase [MNI coordinates, cluster size (k), inverse Bayes factors (BF01) values indicating the evidence in favor of null interaction effects].
Brain region x y z k r(33) p value BF01 vmPFC 0 40 −12 110 −0.088 0.614 4.21 R. amygdala 32 4 −22 175 0.028 0.875 4.69 L. amygdala −26 −4 −20 118 0.088 0.613 4.21 R. caudate/ventral striatum 6 −8 −8 69 0.153 0.379 4.12 L. caudate/ventral striatum −6 −6 −8 52 0.168 0.336 3.72 Correlation between emotion and loss main effects in clusters of our primary ROIs (MNI coordinates, cluster size (k), Pearson's correlation (r) and the corresponding p value, inverse Bayes factors (BF01) values indicating the evidence in favor of null correlation effects).
Outcome condition Mean ± SD Please rate how you felt when the NO-LOSS information was signaled by a POSITIVE image 7.85 ± 1.33 Please rate how you felt when the LOSS information was signaled by a POSITIVE image 4.41 ± 2.14 Please rate how you felt when the NO-LOSS information was signaled by a NEUTRAL image 6.63 ± 1.86 Please rate how you felt when the LOSS information was signaled by a NEUTRAL image 3.89 ± 2.05 Average self-reported pleasantness ratings for the four outcome conditions on a scale of 1–9 (1 being unpleasant and 9 being pleasant), collected at the end of each phase.
Conditions Phase 1 Phase 2 Mean ± SD (%) t(34) p value Mean ± SD (%) t(34) p value Chair choices 49.29 ± 13.69 −0.31 0.76 42.62 ± 14.27 −2.06 0.004 Plane choices 49.37 ± 13.83 −0.27 0.79 55.99 ± 14.59 2.43 0.021 Chair choices that led to Loss outcome 51.03 ± 5.26 1.15 0.26 53.28 ± 10.21 1.90 0.07 Plane choices that led to Loss outcome 48.86 ± 5.59 −1.20 0.24 48.51 ± 6.69 −1.31 0.19 In Phase 1, positive stimulus signaled a loss outcome and neutral stimulus signaled a no-loss outcome. In Phase 2, neutral stimulus signaled a loss outcome, and positive stimulus signaled a no-loss outcome. The t(34) and corresponding p values are based on a one-sample t test against the chance value (50%).
- Table 8.
Average estimated learning rates (α) for the different stimulus types, chair and plane and deemed high and low inherent value stimuli for the two phases separately
Stimulus Phase 1 Phase 2 Mean ± SD (%) t(34) p value Mean ± SD (%) t(34) p value Chair 0.101 ± 0.182 3.264 0.003 0.156 ± 0.268 3.466 0.002 Plane 0.199 ± 0.251 4.692 <0.001 0.214 ± 0.284 4.463 <0.001 High value 0.139 ± 0.201 4.382 <0.001 0.216 ± 0.296 4.384 <0.001 Low value 0.161 ± 0.246 3.628 <0.001 0.154 ± 0.255 3.518 <0.001 In Phase 1, positive stimulus signaled a loss outcome and neutral stimulus signaled a no-loss outcome. In Phase 2, neutral stimulus signaled a loss outcome, and positive stimulus signaled a no-loss outcome. The t(34) and corresponding p values are based on a one-sample t test against zero.
In this issue
