Pupil size asymmetries are modulated by an interaction between attentional load and task experience

In a recently published study [1], we investigated how human pupil sizes are modulated by task experience as well as attentional load in a visuospatial task. In particular, participants performed a multiple object tracking (MOT) task while pupil sizes were recorded using binocular eyetracking measurements. To vary the attentional load, participants performed the MOT task either tracking zero or up to five targets. To manipulate the task experience, participants performed the MOT task on three consecutive days. We found that pupil sizes systematically increased with attentional load and decreased with additional task experience. For all these analyses, we averaged across the pupil sizes for the left and right eye. However, findings of a recent study [2] have suggested that also asymmetries in pupil sizes could be related to attentional processing. Given these findings, we further analyzed our data to investigate to what extent pupil size asymmetries are modulated by attentional load and task experience. We found a significant interaction effect between these two factors. That is, on the first day of the measurements, pupil size asymmetries were not modulated by attentional load while this was the case for the second and third day of the measurements. In particular, for the second and third day, pupil size asymmetries systematically increased with attentional load, indicating that attentional processing also modulates pupil size asymmetries. Given these results, we suggest that an increase in task experience (and associated reductions in arousal) uncover modulations in pupil size asymmetries related to attentional processing that are not observable for typical arousal levels. We suggest that these modulations could be a result of right-lateralized attentional processing in the brain that in turn influences structures involved in the control of pupil sizes such as the locus coeruleus. We can exclude a number of possible alternative explanations for this effect related to our experimental setup. Yet, given the novelty of this finding and the arguably speculative explanation of the underlying mechanisms, we suggest that future studies are needed to replicate the present effect and further investigate the underlying mechanisms.

For the past decade, researchers have investigated how modulations of pupil sizes are 2 related to cognitive processes such as as decision-making [3][4][5][6], attention [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21], 3 emotions [22,23], language [24], and memory [25][26][27][28]. Moreover, changes in pupil sizes 4 have been repeatedly associated with changes in arousal [29][30][31][32]. In these studies, 5 modulations of pupil size were either measured in only one eye or the pupil sizes were 6 averaged across both eyes. However, to the best of our knowledge, only one study has 7 also investigated whether pupil size asymmetries (i.e., differences in pupil sizes between 8 the left and right eye) are related to attentional processing [2]. In this study [2], 9 self-rated assessments of attention ability significantly correlated with pupil size 10 asymmetries, suggesting that also pupil size asymmetries could be modulated by 11 attentional processing. Such a link between pupil size asymmetries and attentional 12 processing could be explained by the right-lateralization of attentional processing in the 13 brain [33][34][35] and the relation of attentional processing to structures involved in the 14 control of pupil sizes (i.e., the locus coeruleus [14,32,[36][37][38]). That is, the 15 right-lateralization of attentional processing could systematically affect structures 16 related to pupil size control which in turn lead to a differential modulation of the left 17 and right pupil sizes. 18 In a recent study [1], we investigated the relation between attentional demands, task 19 experience, and pupil sizes (i.e., averaged across the left and right) in a visuospatial task 20 (i.e., a multiple object tracking (MOT) task). In this study, to vary the attentional load, 21 participants performed the MOT task either tracking zero or up to five targets. To 22 manipulate the task experience, participants performed the MOT task on three 23 consecutive days. We found that pupil sizes systematically increased with attentional 24 load and decreased with additional task experience. However, to date, it has not been 25 investigated whether changes in attentional load also differentially affect pupil size 26 asymmetries. In order to address this question, we further analyzed the data from our 27 previous study to investigate to what extent pupil size asymmetries are related to 28 changes in attentional load. Given that we measured participants on three consecutive 29 days, we also investigated the relation of pupil size asymmetries to task experience and 30 the interaction between the factors attentional load and task experience. 31

32
For details on the methodology of the study, data preprocessing, behavioral 33 performance in the MOT task, and results related to pupil sizes averaged across the left 34 and right eye, we refer to our published manuscript [1]. 35 For investigating pupil size changes that differentially affect the left and right eye, 36 we performed the following normalization on the pre-processed data. We first 37 subtracted the median pupil sizes of the right eye from pupil sizes of the left eye for 38 each trial. Note, we took the median pupil size across three to nine seconds within the 39 tracking period (i.e., while participants tracked the target objects on the computer 40 screen) to avoid perceptual and executive confounds [1]. We then averaged these values 41 for each participant, separately for each number of targets in the MOT task and day. 42 We normalized the pupil size asymmetries for targets one to five by subtracting the 43 pupil size asymmetries for the passive viewing condition and dividing the result by the 44 pupil size in the passive viewing condition averaged across both eyes.

45
A descriptive overview of the normalized pupil size asymmetries can be seen in Fig 1. 46 In this overview, no clear modulations of pupil size asymmetries by attentional load are 47 visible on the first day. On the second and third day, however, pupil size asymmetries 48 appear to be differentially affected by the attentional load conditions. This observation 49 2/8 suggests an interaction effect between attentional load and task experience. were not significant. 66 We followed up this analysis by calculating a separate Pearson correlation coefficient 67 for each participant for each day, correlating the attentional load with the pupil size 68 asymmetries. For this measure, a negative correlation would indicate that the pupil size 69 asymmetries become larger with increasing attentional load. We tested these correlation 70 coefficients against zero using a one sample t-test -for a descriptive overview, see Fig   71   2). Note, prior to entering the correlations in the t-test, we applied a Fisher In sum, we found that pupil size asymmetries between the left and right eye are 79 modulated by the attentional load in a visuospatial task. However, these modulations 80 were small, and only present on the second and third day of the measurements.

81
Moreover, these modulations were not as fine-grained as those observed for the pupil 82 sizes averaged across both eyes [1]. 83 Given that these modulations were only present on the second and third day of the 84 measurements suggests that pupil size asymmetries are additionally influenced by task 85 experience and possibly associated reductions in arousal. That is, larger effects 86 modulating pupil sizes possibly due to arousal could mask smaller modulations related 87 to pupil size asymmetries and attentional processing. Modulations of pupil size 88 asymmetries related to attentional processing may only become visible after effects of 89 arousal are considerably reduced.

90
Typically, asymmetries in pupil sizes are related to pathological conditions and are 91 assumed to be consensual otherwise. In such pathological cases, the magnitude of the 92 asymmetry is considerably larger than reported in the present study. That is, 93 asymmetries are clearly visible when pupil sizes are visually inspected [39,40]. Here, 94 however, the reported data suggests that pupil size asymmetries also could be 95 modulated by an interaction of attentional load and task experience. As pointed out in 96 the introduction section, these modulations could be a result of right-lateralized 97 attentional processing in the brain [33][34][35] that in turn influences structures involved in 98 the control of pupil sizes such as the locus coeruleus [14,32,[36][37][38].

99
As a point of note, one could suggest that modulations of pupil size asymmetries can 100 alternatively be explained by factors related to the setup. One example is that 101 4/8 participants could lean more towards the left during measurements or the left camera 102 could be positioned more away from the eye than the right camera. We did not 103 specifically control for these factors in our setup. However, if these factors would 104 systematically alter results, we likely would already have observed pupil size 105 asymmetries on the first day of the measurements. Yet, pupil size asymmetries were 106 only present on the second and third day of the measurements. Also note, our 107 measurement of pupil size asymmetries is a directional measure of the asymmetries.

108
That is, we always subtracted the pupil size of the right eye from the pupil size of the 109 left eye). We did not use an absolute measure of the asymmetries, i.e., we do not take 110 the absolute value of the calculated differences. We suspect that any factors related to 111 the setup that spuriously could produce pupil size asymmetries are subject to random 112 processes (e.g., as noted above, an individual participant could lean more towards the 113 right while another participant could lean more to the left). These random processes 114 should likely affect absolute measures of the pupil size asymmetries. However, these 115 processes should not systematically affect directional measures of pupil size asymmetries 116 as it is unlikely that a large majority of participants were systematically measured 117 differently on the first day compared to the second and third day of the measurements. 118 Moreover, measurements for the different days for participants were often conduced on 119 overlapping days across participants. For instance, participants who were measured on 120 their second or last day of the measurements were measured on the same day as  [14,32,[36][37][38]. 132 Given these findings, future neurophysiological studies could also investigate the 133 underlying processes that modulate pupil size asymmetries and may relate these to 134 structural asymmetries in the brain [41,42].