Average Blink rate, Blink count, and pupil dilation responses as a function of ADHD indication

We first analyzed average measurements to test group (ADHD, Control) differences in blinking patterns and pupil response during each condition (frequent, sparse, static) and their interaction using three Linear mixed-effects models (Table 2). Average measurements were calculated for each participant: mean eye blink count during paradigm; eye blink rate was calculated as the number of blinks per visible minute (valid eye tracking data); and mean pupil dilation response as the mean change of each trial. Six participants were excluded from the analysis: three due to an extreme blink rate (above 2 blinks per second) and three due to missing pupil data (above 75%). Their removal did not alter group differences in ASRS-V1.1 scores, age, gender, or place of residence.

Using global measures, there were no main effects of group or condition, nor were there significant interactions for blink rate. Similarly, no main effects or interactions were observed for mean blink count, and group differences were nonsignificant in all three conditions. For the mean pupil dilation response, the group effect was nonsignificant, and no interactions emerged; however, a main effect of condition indicated a smaller pupil size in the static condition compared to the frequent condition (p < .001). ADHD status was not associated with differences in average blink measures or pupil dilation, though pupil size was generally least dilated in the static condition, irrespective of group.

Table 2 Comparison of Blink rate, Blink count, and pupil dilation response between participants with and without ADHD-ind across conditions (static, sparse, frequent)The effect of simulated blinking and ADHD indication on contingent blinking response

To examine whether the blinking rate of participants changed in response to simulated blinking appearing in the video, we calculated the mean Z score of blink rate in each 250-ms bin between − 500 and 1250 ms around each simulated blink. A difference in rate occurring in the relevant timing interval of the simulated blink event was referred to as blink synchrony onset. As seen in Fig. 2, participants in the control group displayed evidence of blink synchrony in both the frequent and sparse video conditions. Specifically, one-sample t-tests against zero (FDR-corrected) indicated that in the frequent condition, participants in the control group showed a significant decrease in blink rate at the exact onset of the simulated blink (0 ms), t(42) = − 3.03, p = .004, FDR-adjusted p = .029. In the sparse condition, control participants also exhibited reduced blink rates just before the simulated blink at − 250 ms, t(42) = − 2.83, p = .007, FDR-adjusted p = .049, with a trend toward reduced blinking at 750 ms, t(42) = − 2.33, p = .025, FDR-adjusted p = .086. In contrast, ADHD-Ind participants did not show reliable synchrony effects after FDR correction. Although uncorrected results suggested reduced blink rates around − 250 ms (p = .021) and 0 ms (p = .22) in the frequent condition, these effects did not survive FDR adjustment (padj = 0.076). No other bins reached significance in the ADHD-Ind group across conditions. Together, these results suggest that blink synchrony to simulated social cues emerges in the control group—particularly around the onset of simulated blinks—. In contrast, synchrony effects are attenuated or absent in individuals with ADHD indications.

Fig. 2

Blink synchrony relative to surrogate data across groups and conditions. Mean Z-scores of blink rate are plotted in 250-ms bins from − 500 to 1000 ms around simulated blink onset (0 ms), separately for the control (top panels) and ADHD-Ind (bottom panels) groups, in frequent (left) and sparse (right) conditions. Error bars represent ± 1 SEM. Asterisks indicate bins where one-sample t-tests against zero were significant after FDR correction. Control participants showed significant reductions in blink rate at 0 ms in the frequent condition and at − 250 ms in the sparse condition, whereas no bins reached significance in the ADHD-Ind group

The effect of simulated blinking and ADHD indication on pupil dilation response

We analyzed pupil dilation response from − 500 to 1250 ms around each simulated blink to examine the effect of simulated blinks on pupil dilation response. Each simulated blink onset (when the eyes began to close) was used as an event representing time 0. Pupil data was aggregated continuously around each event for each subject individually, creating a mean continuous response for each subject. An independent sample t-test was calculated for each time frame between participants with ADHD-Ind and control in each condition (sparse, frequent). Welch t-tests showed no differences in the frequent condition at any time bin. In contrast, the sparse condition exhibited a region of significant between-group differences from − 80 to 560 ms, with ADHD-Ind showing a constricted pupil response compared to control [t(36.77) = 2.90, p = .006, 95% CI 0.029, 0.162] (Fig. 3).

Fig. 3

Pupil dilation per condition (sparse vs. frequent). Baseline-corrected pupil change (mm) from − 500 to 1250 ms around simulated-blink onset (0 ms), shown separately for frequent (top) and sparse (bottom) conditions. Lines are group means (Control = red, ADHD-Ind = teal) with shaded 95% CIs; the dashed line marks zero. The bottom rug marks point-wise between-group Welch t-tests at each time bin (green = contiguous bins with p < .1). No between-group differences appear in the Frequent condition. In contrast, the Sparse condition shows a sustained window of group differences from approximately − 80 to 560 ms (ADHD-Ind more negative/constricted than control)

A one-sample time-course test within groups was performed to examine the different responses to the condition. Results indicated that participants in the control group did not differ from zero at any bin, whereas participants with ADHD-Ind showed a window of significant difference from − 40ms to 400ms.

Within the ADHD group, the pupil response to sparse blinks was significantly less dilated than its response to frequent blinking when averaged over − 40 to 400 ms [t(22.99) = 2.52, p = .019, 95% CI 0.022, 0.225], indicating greater pupil dilation around blink presentation to frequent simulated blinks (Fig. 4)—a pattern more similar to that of controls in this condition.

Fig. 4

Pupil dilation per group (ADHD-Ind vs. Control). Baseline-corrected pupil change (mm) over the same epoch, faceted by Group (Control top, ADHD-Ind bottom). Lines are condition means (Frequent = red, Sparse = teal) with 95% CIs; the dashed line marks zero. The bottom rug marks point-wise within-group frequent vs. sparse Welch t-tests (green = contiguous bins with p < .1). In ADHD-Ind, the sparse response is significantly more negative than frequent over roughly − 40 to 400 ms; no sustained Frequent–Sparse differences are evident in the Control group

Overall, these results show a condition-specific group difference in participants with ADHD-Ind during sparse blinks, while frequent blinks elicit similar responses across groups.

The effect of simulated blinking and ADHD symptomatology on contingent blinking response

To better understand the relations between ADHD symptomatology and synchronized blinking response, we analyzed the relations between participants’ synchronized blinking (represented by their Z score), the condition (frequent, sparse), time, and their ASRS-V1.1 total score. For that purpose, we fitted a linear mixed-effects model to standardized blink synchrony.

$$ \:Z\sim \:}\:} \times \:} \times \:}\left( }\left| }} \right.} \right) $$

To explore the triple interaction between ADHD symptom severity, time and blink rate conditions, the number of comparisons tested was reduced to four bins around the simulated blink. Results showed a significant negative main effect for ADHD symptomatology, such that higher ADHD symptom scores predicted lower blink synchrony (\(\:b=-0.224\), \(\:SE=0.094\), \(\:t\left(352\right)=-2.37\), \(\:p=.018\), 95% CI [− 0.409, − 0.039]). Neither the time (\(\:b=0.160\), \(\:SE=0.146\), \(\:t\left(352\right)=1.09\), \(\:p=.275\)), nor the condition (sparse, frequent) (\(\:b=0.102\), \(\:SE=0.133\), \(\:t\left(352\right)=0.77\), \(\:p=.444\)), reached significance. All interactions were insignificant (ADHD symptomatology × time: \(\:p=.111\); ADHD symptomatology × Condition: \(\:p=.402\); time × Condition: \(\:p=.951\); ADHD symptomatology × time × condition: \(\:p=.482\)), indicating the effect of ADHD symptomatology was fairly constant across time and conditions. Semi-partial \(\:^\)estimates indicated that the fixed effects jointly explained 3.3% of the variance in blink synchrony (95% CI [1.8%, 9.8%]). The unique contribution of ADHD symptomatology was 1.7% (95% CI [0.1%, 5.4%]), a small significant effect.

The effect of simulated blinking and ADHD symptomatology on contingent pupil dilation response

To examine how ADHD symptomatology relates to the pupil-dilation response, we modeled baseline-corrected pupil change across time and condition using a linear mixed-effects model.

$$ \begin \:Pupil\:Dilation\sim \: & }\:} \times \:(}^ ) \\ & \times \:}(}\:}). \\ \end $$

To test the triple intecration effect on pupil dilation, we used pupil response from − 500 to 1250 ms around the simulated blink. Results show a change across time after the simulated blink (Time: b = 0.0055, SE = 0.0010, t(≈ 30,000) = 5.68, p < .001; Time²: b = 0.0048, SE = 0.0011, t(≈ 30,000) = 4.42, p < .001), and a lower pupil dilation response in sparse condition relative to frequent (b = − 0.0271, SE = 0.0026, t = − 10.46, p < .001). The main effect of ADHD symptomatology was not significant (b = − 0.0269, SE = 0.0185, p = .151); however, interactions indicated ASRS-dependent differences in trajectory and condition: ADHD symptomatology × Time (b = − 0.00283, SE = 0.00097, p = .0036), ADHD symptomatology × Time² (b = 0.00362, SE = 0.00109, p = .00093), ASRS×Sparse (b = 0.02836, SE = 0.00257, p < .001), Time² × Condition-sparse (b = 0.01412, SE = 0.00193, p < .001), and ADHD symptomatology × Time² × Condition-sparse (b = − 0.00432, SE = 0.00192, p = .025); the Time × Condition-sparse interaction was not significant (p = .112). Semi-partial \(\:^\:\)estimates showed the fixed effects explained 3.0% of variance in pupil dilation, and the unique contribution of ADHD symptomatology was 1.1%, a small significant effect.

Overall, the Results suggest that, unlike global measures of average blink or pupil responses, ADHD-related differences in automatic neural activity emerge in time-locked responses, which are vital in social encounters. The control group showed clear blink synchrony to the social simulated stimuli, whereas the ADHD-Ind group exhibited attenuated synchrony and greater pupil constriction, specifically in the sparse condition. Higher ADHD symptoms were associated with lower synchrony and subtle changes in the pupil dilation time course. Overall, the pattern points to ADHD related timing-specific alterations in alignment to social cues (especially sparse blinks), rather than baseline differences.