Self-association enhances early attentional selection through automatic prioritization of socially salient signals
- University of Aberdeen, United Kingdom
- Durham University, United Kingdom
- Philipp University of Marburg, Germany
Figures
Figure 1
Mechanisms of attentional selection.
Attentional selection can occur via different mechanisms: changes in the allocation or the availability and allocation of attentional resources. Panel A shows an example stimulus display with two identical perceptual objects (gray hexagons) on a dark background, and the processing resources (, blue columns) that are distributed equally across these stimuli. In this case, the attentional weights for the reference (; left stimulus) and probe (; right stimulus) are identical. The processing rates for the reference () and probe () stimuli are given as the processing resources that are allocated to each of the two stimuli, i.e. . Panels B and C show example stimulus displays with two perceptual objects, where one has a higher perceptual salience (luminance and color contrast). Panel B indicates a mechanism whereby the same processing resources are distributed differentially across the stimulus display, with more resources being given to the more salient stimulus. This is reflected in a change of relative weight, with constant processing capacity. Panel C indicates a mechanism whereby the amount of processing resources increases, along with a differential distribution of these resources. To arbitrate between these two mechanisms, we employed model comparisons to assess whether changes in relative attentional weight or changes in absolute processing rates (capacity and weights) better explained experimental data.
Figure 2
Decision dimensions.
Dissociating the processing stages at which social association may affect early attentional selection via different decisional dimensions. Automatic effects of self-association would assume that active decoding of the associated social identity is not necessary. In this case, the social identity associated with a specific perceptual feature renders this feature more salient, without having to be consciously recalled. On the other hand, some studies suggested that the self needs to be a decisional criterion. In this case, self-prioritization effects in attentional selection would require active decoding of the social associations. Altering the decisional dimension (asking which shape vs whose shape), without shifting attention from the crucial perceptual feature (shape), allows disentangling these processes. Note that the directionality of the sensory and social information does not make assumptions about the temporal dynamics of the underlying process.
Figure 3
Task design.
(a) Temporal order judgement task (TOJ) design. Following an initial presentation of the complete stimulus array, target shapes, which were relatively larger in size compared to background shapes, flickered with a variable stimulus onset asynchrony that was systematically varied between -/+ 83 ms with a higher presentation frequency at small SOAs. After the stimulus presentation, participants had to indicate which of the two shapes flickered first by selecting the correct shape (baseline conditions, perceptual salience conditions, social salience condition with perceptual decision boundary), or the identity label of the shape-associated social identity (social salience condition with social decision boundary). Stimulus displays consisted of two types of colored shapes (perceptual objects), distributed across two hemifields in an 8 x 8 grid. Targets would appear on each side at either of the four central locations. Lateralization of the specific perceptual objects was randomized across trials. (b) Perceptual matching task design. Participants associated one of the two shapes with themselves, and one with another, anonymous participant. Associations between social identities and perceptual objects were counterbalanced across participants. Pairs of shapes and social identity labels were presented on screen. These could either be congruent (matching) or incongruent (mismatching). Participants had to respond whether the pair matched in the learned association or mismatched. Location of the shapes and labels (above, below fixation) was counterbalanced across the task. (c) Task structures for Experiments 1 and 2. Both experiments began with a TOJ baseline task. Experiment 1 utilized non-salient targets exclusively, while Experiment 2 included both perceptually salient and non-salient targets. These were presented in randomly intermixed order. Next, targets were associated with social identities through a matching task. Following this association learning phase, which establishes social salience in the shapes, participants completed the same TOJ task again. In Experiment 1, they completed one block using a social decision dimension and one block using a perceptual decision dimension. The order of these blocks was counterbalanced across participants to reduce the influence of order effects in the results. In Experiment 2, perceptually salient and non-salient stimuli were presented in an intermixed fashion, and participants responded within the social decision dimension. Each task block was preceded by 8 (matching) to 14 (TOJ) practice trials.
Figure 4 with 1 supplement
Model structure and cross-experimental social salience effects.
Hierarchical model structure shows how the parameters of interest (, , , and ) were estimated from within-participant differences between the neutral baseline (gray) and social salience (blue) condition. The better model is depicted, in which processing capacity was estimated for each condition separately, suggesting that changes in absolute processing rates, rather than relative attentional weights, have been underlying attentional selection effects of social salience. Mathematical formalization of the relation between the model nodes is given on the right. Density plots indicate the highest density estimates for the different processing parameters of interest. Neutral baseline parameters are given in absolute parameter values, with processing capacity shown as items/ms and the relative attentional weight for the probe (a shape that was subsequently associated with the self). Social salience parameters are shown in change scores, relative to baseline, depicting an increase and decrease in processing capacity and attentional weight, respectively. Absolute processing rate changes for the probe (self-associated) and reference (other-associated) shapes, as well as their relative change, are shown on the bottom right.
Figure 4—figure supplement 1
Individual estimates: Absolute processing rates () for the probe stimulus (self-associated), shown for individual participants for the social baseline (gray; ) and the social salience condition in which the identity had to be reported (dark blue; ).
Points show the estimated posterior means, thick lines indicate the central quartiles, and thin lines indicate the 95 highest density interval. Processing rates are given in items/ms.
Figure 5 with 2 supplements
Social salience effects with different decision dimensions.
Hierarchical model structure shows how social salience effects were estimated between social salience conditions (blue: social decision dimension; turquoise: perceptual decision dimension) and the neutral baseline condition (gray; see Figure 4 caption for details and formalizations) in Experiment 1. The right plot shows how relative processing rates were calculated, at the individual participant level, from social salience-induced processing rate changes for the self-associated and other-associated shapes. Density plots indicate the group-level highest density intervals for the processing capacity and absolute processing rate estimates, given in items/ms. Additionally, the 95% HDIs are presented alongside the group means. The relative change in processing rates () can be interpreted directly as the processing rate advantage of the self-associated over the other-associated stimuli. Raw response data and parameter estimates for individual participants are provided in Figure 5—figure supplement 1 and 2, respectively.
Figure 5—figure supplement 1
Individual psychometric functions: Individual participant response data indicating the proportion with which participants responded that the probe flickered first as a function of stimulus onset (flicker) asynchrony.
Different conditions are shown in different shadings. The panel on the right shows an example participant with a group-representative response pattern: an increase of ‘probe first’ responses when the probe was self-associated and the shape of the stimulus had to be reported. This pattern is specifically visible at low onset asynchronies. Furthermore, this participant shows a decreased proportion of ‘probe first’ responses when the probe was self-associated and the social identity had to be reported.
Figure 5—figure supplement 2
Individual estimates: absolute processing rates () for the probe stimulus (self-associated), shown for individual participants in Experiment 1.
Different colors indicate different conditions: the baseline (gray; ), the social salience condition in which the shape had to be reported (light blue; ) and the social salience condition in which the identity had to be reported (dark blue; ). Points show the estimated posterior means, thick lines indicate the central quartiles, and thin lines indicate the 95 highest density interval. Processing rates are given in items/ms.
Figure 6 with 2 supplements
Social and perceptual salience effects.
Hierarchical model structure shows how social salience only (blue) and perceptual salience only (orange) effects were estimated relative to their respective neutral baseline conditions (gray; see Figure 4 caption for formalizations and main text for details) in Experiment 2. To assess processing parameters for the interaction of social and perceptual salience (purple, pink), we calculated change scores, indicative of perceptual salience effects, relative to the perceptual neutral baseline. This can be interpreted as the effect of perceptual salience that is present when shapes are associated with the own or another social identity. Density plots indicate the group-level highest density intervals for the processing capacity and processing speed estimates, given in items/ms. Additionally, the 95% HDIs are presented alongside the group means. The relative speed change () can be interpreted directly as the processing rate advantage of the self-associated over the other-associated stimuli (social salience only), or the perceptually salient over the perceptually non-salient stimuli (perceptual salience only, social + perceptual salience). Raw response data and parameter estimates for individual participants are provided in Figure 5—figure supplements 1 and 2, respectively.
Figure 6—figure supplement 1
Individual psychometric functions: Individual participant response data indicating the proportion with which participants responded that the probe flickered first as a function of stimulus onset (flicker) asynchrony.
Different conditions are shown in different shadings.
Figure 6—figure supplement 2
Individual estimates: Absolute processing rates () for the probe stimulus (perceptually salient), shown for individual participants for the baseline (gray; ), for the mere perceptual salience condition (orange; ), for the perceptual salience condition in which the probe was self-associated (purple; ), and the perceptual salience condition in which the probe was other-associated (pink; ).
Points show the estimated posterior means, thick lines indicate the central quartiles, and thin lines indicate the 95 highest density interval. Processing rates are given in items/ms.
Figure 7 with 1 supplement
Interaction of social and perceptual salience.
Interaction effects of social and perceptual salience on processing rates. Probability density plots of processing rate change parameters when the stimulus was perceptually salient and self-associated (left panel) or other-associated (middle panel). The right panel shows the difference in perceptual-salience induced processing benefit between the self- and other-associated stimuli.
Figure 7—figure supplement 1
Formalization of interaction: formalization of interaction effect assessment, using attentional weights.
Note that, instead of attentional weights, we report processing rates. However, the same effects that are reported in the main text are reproduced when attentional weights are used in the analysis.
Figure 8
Self-prioritization in matching.
Matching task results showing robust self-prioritization effects in both experiments, indicated by the enhanced accuracy towards self-associated information compared to other-associated information on match-trials. Colored points indicate individual participants. Black superimposed points and error bands indicate group means and .
Figure 9
Individual differences reveal how task parameters are related.
Scatter plots for Experiments 1 and 2 showing individual self-prioritization effect benefits in the shape-label matching task predicting socially induced changes in attentional processing rates for the self-associated/salient (positive) or other-associated/non-salient (negative) stimuli. Solid lines indicate linear best fit. The right scatter plot shows the changes in individual, absolute processing rates in response to social versus perceptual decision judgements, for both experiments. Experiments and respective best linear prediction lines are color-coded in darker (Experiment 1) and lighter (Experiment 2) gray, to allow distinguishing between conditions that used social salience (dark gray) or perceptual salience (light gray) with the perceptual decision dimension. Bayes factors assessing the probability of a linear correlation and posterior estimation info is provided above each plot.
Figure 10
Parameter estimates and their respective uncertainties for social and perceptual salience.
Figure 11
Estimated power for observing a reliable self-bias in attentional selection for the perceptual salience (orange solid line) and social salience (green dashed line) conditions.
Shaded bands indicate their respective 95% highest density intervals.
Tables
Table 1
Posterior coefficient summaries for perceptual salience and other-association.
| Coefficient | P(incl) | P(incl|Data) | Mean | SD | Lower | Upper | |
|---|---|---|---|---|---|---|---|
| Intercept | 1.000 | 1.000 | 1.0 | 0.052 | 0.011 | 0.031 | 0.073 |
| 0.556 | 1.000 | 4638.74 | 0.833 | 0.200 | 0.434 | 1.232 | |
| 0.556 | 0.779 | 2.83 | –1.191 | 0.624 | –2.435 | 0.053 | |
| * | 0.333 | 0.414 | 1.42 | 6.081 | 6.667 | –7.216 | 19.378 |
Table 2
Posterior coefficient summaries for perceptual salience and self-association.
| Coefficient | P(incl) | P(incl|Data) | Mean | SD | Lower | Upper | |
|---|---|---|---|---|---|---|---|
| Intercept | 1.000 | 1.000 | 1.0 | 0.021 | 0.006 | 0.008 | 0.034 |
| 0.556 | 0.995 | 153.25 | 0.405 | 0.119 | 0.167 | 0.643 | |
| 0.556 | 1.000 | 2458.52 | 0.630 | 0.372 | –0.111 | 1.372 | |
| * | 0.333 | 0.573 | 2.68 | 4.323 | 3.976 | –3.606 | 12.252 |
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Self-association enhances early attentional selection through automatic prioritization of socially salient signals
eLife 13:RP100932.
https://doi.org/10.7554/eLife.100932.3
