Do pupillary responses during authentic slot machine use reflect arousal or screen luminance fluctuations? A proof-of-concept study.

Modern slot machines are among the more harmful forms of gambling. Psychophysiological measures may provide a window into mental processes that underpin these harms. Here we investigated pupil dilation derived from eye tracking as a means of capturing changes in sympathetic nervous system arousal following outcomes on a real slot machine. We hypothesized that positively reinforcing slot machine outcomes would be associated with increases in arousal, reflected in larger pupil diameter. We further examined the contribution of game luminance fluctuations on pupil diameter. In Experiment 1A, experienced slot machine gamblers (N = 53) played a commercially-available slot machine in a laboratory for 20 minutes while wearing mobile eye tracking glasses. Analyses differentiated loss outcomes, wins, losses-disguised-as-wins, and (free-spin) bonus features. Bonus features were associated with rapid increases in pupil diameter following the onset of outcome-related audiovisual feedback, relative to losses. In Experiment 1B, luminance data were extracted from captured screen videos (derived from Experiment 1A) to characterize on-screen luminance changes that could modulate pupil diameter. Bonus features and wins were associated with pronounced and complex fluctuations in screen luminance (≈50 L and ≈25L, respectively). However, the pupil dilation that was observed to bonus features in Experiment 1A coincided temporally with only negligible changes in screen luminance, providing partial evidence that the pupil dilation to bonus features may be due to arousal. In Experiment 2, 12 participants viewed pairs of stimuli (scrambled slot machine images) at luminance difference thresholds of ≈25L, ≈50L, and ≈100L. Scrambled images presented at luminance differences of ≈25L and greater were sufficient to cause pupillary responses. Overall, pupillometry may detect event-related changes in sympathetic nervous system arousal following gambling outcomes, but researchers must pay careful attention to substantial in-game luminance changes that may confound arousal-based interpretations.

Introduction

Eye tracking technology is becoming more widely used in gambling research [1-4], and equipment costs have dropped substantially in recent years. In addition to measuring overt visual attention, these developments have enabled psychophysiological measures involving the eye’s pupil. Phasic changes in pupil diameter can indicate psychological reactions to internal or external events [5-7]. This study explores the acquisition and interpretation of pupillary data obtained during real-world gambling, in the specific context of authentic slot machine games.

The recording of psychophysiological measures during gambling may provide a window into the psychological processes involved in gambling (e.g., positive reinforcement, excitement) and may further uncover individual differences relevant to the risk of experiencing gambling-related harms. Beginning in the 1980s (e.g., [8]), a range of physiological measures have been examined in relation to gambling, including various cardiographic indices [9-12], electrodermal indices [13-16], muscular indices [17-19], and body temperature [20]. Some of these measures have been further linked to problem gambling status, although not consistently so [21-23]. From a methodological perspective, physiological measures vary in several respects, including the speed of response (enabling the detection of ‘event-related’ responses to gambling wins or losses) and the specificity of any physiological interpretation (e.g., to sympathetic vs parasympathetic nervous system activity; [24, 25]).

The present study focuses on physiological signals during slot machine gambling, one of the more harmful forms of gambling [26-28]. Modern slot machines are highly engineered games, and the harms associated with their use have been linked to a number of ‘structural characteristics’ including fast game speed [29], regular reinforcement [30], and intense audiovisual feedback [31, 32]. Traditional perspectives on gambling motives and the nature of gambling reinforcement have emphasized the role of excitement (i.e., thrill or ‘rush’) [33-35]. But recent work has also linked the problematic use of slot machines to a state of immersion, aligned with motives of coping and escape [36, 37], and notably, excitement and immersion could be hypothesized to have quite distinct psychophysiological signatures (e.g., [12]).

To date, psychophysiological studies on slot machine gambling have primarily examined ‘tonic’ changes, occurring over several minutes of play [22, 38–40]—often with ambiguous results [41]. In terms of event-related (or ‘phasic’) analyses, changes in skin conductance have been described following wins and near-miss outcomes, compared to losses [13, 42, 43], but these studies have used rather simplified slot machine ‘simulations’. As skin conductance levels peak around 4—5s post-stimulus, this physiological measure is still limited in its suitability for use with modern, authentic slot machines, on which successive bets can often be placed every 3–4 seconds [44]. Thus, alternative psychophysiological measures that display a faster response profile could have considerable value in characterizing gambling arousal changes in relation to specific in-game events.

The pupil dilator muscle reacts within 200–2,500 milliseconds of a stimulus [45], a property that is well suited to ‘event related’ assessments. Pupillary responses in consistently-lit environments indicate activation of the sympathetic nervous system [46-49], and are further modulated by noradrenergic [50] and dopaminergic [51] neurotransmission. Pupillary measures are increasingly studied within research on decision-making, showing sensitivity to key components of gambling including choice, uncertainty, reward anticipation, and reward receipt [52-57]. For example, in a study using a two-choice lottery task [31], changes in pupil diameter during the choice and anticipation intervals were associated with both the size of the available reward and its probability, and pupil dilation was further increased by the addition of audiovisual feedback to the task.

The overarching aim of the present study was to examine pupillary changes in the context of modern, authentic slot machine gambling. Experiment 1A analyzed pupil diameter from an eye tracking experiment [58] in which experienced slot machine gamblers played a genuine slot machine game for 20 minutes. We distinguished four types of gambling outcomes for assessing event-related changes in pupil diameter. In addition to standard ‘wins’ (where the payout exceeds the bet) and ‘losses’ (i.e., zero payout outcomes), we also coded losses-disguised-as-wins (LDWs) [59] as celebrated outcomes where the amount won is less than the bet; and free-spin bonus features [60, 61], as rare events that are accompanied by distinctive and intense audiovisual cues, and are known to be highly appealing outcomes to regular slot machine gamblers [62]. We hypothesized that these three types of reinforcing outcomes would be associated with significant pupil dilation (i.e., increased pupil diameter) relative to losses.

Experiment 1A

Methods

Participants and procedure

This paper reports pupillary analyses of an eye tracking experiment reported in [58], involving 53 community-recruited slot machine gamblers (32 men and 21 women; mean age = 33.53, SD = 12.30). Participants were recruited through craigslist.ca advertisements. They were at least 19 years old and reported playing a slot machine in the past 12 months. All participants scored < 8 (for high risk of gambling problems) on the Problem Gambling Severity Index; [63]). They had no history of neuropsychiatric or ophthalmic disease, psychotropic medication use, or recent/severe traumatic brain injury. They had normal or corrected-to-normal visual acuity with prescription strength between -4 and +4 diopters.

Participants were paid $20 CAD for their participation and received any earnings on the slot machine task as a bonus (up to $20). All experimental protocols were approved by UBC’s Behavioural Research Ethics Board, and participants provided written informed consent.

Apparatus. Participants gambled using an endowment of $40 cash on a genuine slot machine (“Buffalo Spirit,” Scientific Games Co., Las Vegas, NV) for 20 minutes, or until they ran out of credit (see S1 Text for additional procedure details). Participants initiated each spin (akin to a ‘trial’ in a psychology experiment) by pressing the bet button on the right-hand side of the slot machine fascia. The reels spun for up to 6 seconds [44] and stopped to reveal one of four possible outcomes: a loss, win, loss-disguised-as-a-win, or a bonus round. Following each spin, any payout and associated audiovisual feedback was presented. Losses have no positively-reinforcing payouts or audiovisuals, and are the most common outcome. Wins and losses-disguised-as-wins involve both a payout and audiovisual feedback, but on losses-disguised-as-wins, the amount won is less than the amount wagered [59, 64]. During audiovisual feedback for wins and losses-disguised-as-wins, the value displayed on the slot machine’s credit and win windows incremented over several seconds, and therefore the duration of the feedback interval to these outcomes is variable and related to the size of the win. Furthermore, the exact value of a given win, loss-disguised-as-a-win, or bonus feature was only known to participants at the end of any outcome-related audiovisual feedback. After the audiovisuals finished, the device became idle, and waited for the participant to initiate the next spin. We defined this epoch as the inter-trial interval.

The free spin bonus feature was awarded by three special (and over-sized) ‘bonus’ symbols visible on the screen by the end of the spin (see Fig 3). This triggered 15 or more free spins that proceeded automatically (i.e., without the player being required to press the bet button). The free spin bonuses doubled the value of any wins and had a unique music track for the duration of the feature. On a bonus feature, any winnings accrued were transferred to the participant’s credit balance only after the last free spin had ended. Because of their uniqueness and lack of participant intervention, we treated the multiple spins within a bonus feature as a single entity.

During the slot machine gambling session, participants wore a pair of mobile eye tracking glasses that provided real-time, natural gaze behavior from both eyes at a rate of 60 Hz (SMI, Teltow, Germany). Eye position data were calibrated using three corner symbols on the game screen and recorded to a Samsung Galaxy Note 4 affixed to the back of the participant’s seat. Participants were able to sit at a comfortable distance from the game; data could be extracted from the eye tracking glasses without requiring participants’ heads to be stabilized.

Data analysis

Data were pre-processed using SMI’s proprietary analysis software (BeGaze 3.7). A binocular average pupil diameter in millimeters (mm) was extracted at 60 Hz for the duration of the slot machine session. To link the pupil data with slot machine outcomes, we derived a time series of game events from a video capture of the slot machine screen, using an internal video splitter and image recognition software that we developed (see S1 Text).

To calculate pupillary responses, we defined three epochs on each spin (see Fig 1): 1) a Baseline epoch was calculated as the mean pupil diameter during the last 200 ms of the reel spin, 2) an audiovisual feedback epoch from 200 ms to 2,500 ms [45] after the reels stopped (T1; this epoch captures the response to the initial audiovisual feedback), and 3) an epoch from 200 ms to 2,500 ms after the offset of the audiovisual feedback (T2; this epoch captures the response to the final amount won). We were interested in both the start (T1) and end (T2) of audiovisual feedback because T1 indexed the first instance when gamblers became aware that reinforcement would be delivered, and T2 indexed their response to finding out the actual monetary value of the reinforcement. If the duration of the feedback was under 2.5s, there was some degree of overlap in the T1 and T2 epochs. This overlap varied by outcome type, with LDWs having the shortest average feedback duration and therefore the most overlap. Loss outcomes have no associated audiovisual feedback, and thus the T2 offset epoch cannot be specified.

Fig 1

Recording epochs for pupillary responses.

Note that on Loss outcomes, there is no ensuing audiovisual feedback and therefore T2 cannot be modelled separately from T1. On the other outcomes, if the duration of audiovisual feedback was less than 2.5 seconds, the T1 and T2 epochs would overlap; see S1 Table for average durations.

In order to measure changes in pupil diameter, we calculated the peak pupil dilation values within the T1 and T2 epochs and calculated the percentage change from the average diameter recorded during the Baseline epoch for that trial [45]. Missing pupil diameter data due to blinks and saccades were estimated using linear interpolation. If more than 18% of samples in a given epoch were missing due to blinks or saccades, it was discarded, as per published recommendations [45]. We excluded 43 outlying values that were more than three standard deviations from the mean response, likely due to equipment error. In total, 5,934 values were included. The resulting distribution of data appeared normal.

A fixed-effects regression model was created to predict pupil diameter for the three positively-reinforced outcome types relative to losses [44, 65, 66]. We fitted three dummy-coded outcome types (wins, losses-disguised-as-wins, and bonus features), using losses as the reference category. We also included predictors representing trial number and credit balance (i.e., credits held at the start of the current spin). Trial number was included to examine changes in pupil magnitude over the duration of the task. The ‘credit balance’ variable was included to account for any variance that the broader financial context (the participant’s ‘performance’) could have on their responses to discrete outcomes.

For transparency and reproducibility, our data and analysis scripts have been archived at https://doi.org/10.5683/SP3/MQVSP3. Please refer to the S1 Text for additional details.

Results

Descriptive data on the number of trials per outcome type are presented in the S1 Text and S1 Table. Fig 2A shows changes of pupil dilation as a function of different time epochs and audiovisual events for a model participant. Although pupil diameter was relatively invariant for losses and losses-disguised-as-wins, we observed event-related changes in pupil diameter for wins and free spin bonus features. Inferential analyses showed that, relative to loss trials, free spin bonus features were associated with a significant pupillary response in the T1 epoch following the onset of audiovisual feedback (B = 3.18, t(5830) = 3.83, p < .001, Table 1, Fig 2C). Wins (B = 0.09, t(5830) 0.43, p = .668) and losses-disguised-as-wins (B = -0.20, t(5830) = -0.96, p = .338) did not differ from losses at T1. Following the offset of audiovisual feedback (T2 epoch), we saw a slight decrease in pupil diameter for wins relative to losses (B = -1.09, t(5830) = -4.99, p < .001), and a larger decrease for bonus features (B = -3.24, t(5830) = -3.97, p < .001). There was no significant effect at T2 for losses-disguised-as-wins (B = -0.25, t(5830) = -1.22, p = .223). Trial number was negatively related to pupillary response magnitude (B > -0.01, t(5830) = -3.32, p = .001, Table 1), indicating that event-related increases in pupil diameter diminished over the course of the session. Participants’ credit balance was not a significant predictor of pupil size (B > -0.01, t(5830) = -1.71, p = .088). Overall, the model accounted for 55% of the variance in task-related pupillary responses (R2adj = .55).

Fig 2

Change in pupil diameter by outcome type in a representative participant and the overall group.

(A) Pupil diameter trace for a representative participant through Baseline, and at the start (T1), and end (T2) of audiovisual feedback. Lines represent the different outcome types. Loss trials have no T2. (B) Mean pupil diameter for all participants. Lines represent the mean of different outcome types. Shaded areas represent the standard error between subjects for that point in time. (C) Predicted pupillary response to different slot machine outcomes, as percentage change from Baseline, minus the response for loss outcomes. Asterisks represent the effect of the outcome relative to the reference category, loss trials. Bars represent the 95% confidence interval. LDW = Loss-Disguised-as-a-Win, * p < .05, ** p < .01, *** p < .001.

Table 1

Event-related change in pupil diameter as a percentage of baseline by trial number, prior credits, outcome type, and outcome phase.

Factor	B	95% CI	t(5830)	p
Trial Number(1)	>-0.01	[-0.01, >-0.01]	-3.32	.001***
Credit Balance(4,000)	>-0.01	[>-0.01, <0.01]	-1.71	.088
Win T1(Loss)	0.09	[-0.33, 0.52]	0.43	.668
T2(Loss)	-1.09	[-1.51, -0.66]	-4.99	< .001***
LDW T1(Loss)	-0.20	[-0.60, 0.21]	-0.96	.338
T2(Loss)	-0.25	[-0.66, 0.15]	-1.22	.223
Bonus T1(Loss)	3.18	[1.55, 4.80]	3.83	< .001***
T2(Loss)	-3.24	[-4.84, -1.64]	-3.97	< .001***

Coefficients (B) represent percentage change in pupil diameter per unit increase of the factor. Subscripted values represent the default value for each factor. LDW = Loss-Disguised-as-a-Win.

* p < .050,

** p < .010,

*** p < .001.

Experiment 1A discussion

We investigated pupillary responses as a potential physiological marker of reward processing among experienced gamblers using an authentic slot machine. We created a regression model that accounted for 55% of the variance in pupillary responses. Pupil diameter increased by approximately 3.2% in response to the onset of audiovisual feedback on free-spin bonus features, relative to loss trials, consistent with sympathetic nervous system activation. For comparison against another appetitive task, similar effects were observed in healthy participants consuming chocolate [45]. We detected no significant increases in pupil diameter to the equivalent epoch following wins and losses-disguised-as-wins; the pupillary response on these outcomes was generally flat (Fig 2B), and their associated confidence intervals considerably overlapped the pupillary activity following losses (Fig 2C). We also observed unexpected decreases in pupil diameter (i.e., pupil contraction) following the offset of audiovisual feedback (the T2 epoch) to bonus features and wins (see S1 Text for further discussion).

Experiment 1B

Game-related changes in screen luminance could also drive changes in pupil diameter, given the established effects of screen brightness on the pupillary light reflex [7]. In visual psychophysics, researchers pay careful attention to stimulus luminance in designing their tasks, often keeping luminance constant to avoid confounding the effect of interest with changes in luminance. Naturally, there is no such reason for slot machine designers to intentionally keep luminance fixed. If bonus features, for example, are accompanied by significant decreases in screen luminance, this may indicate that the pupil dilation observed in Experiment 1A is not mediated by physiological arousal and/or excitement but instead driven by luminance changes. By the same reasoning, any systematic luminance-related increases on wins and LDWs could elicit pupillary contraction that could conceivably cancel out any arousal-related pupil dilation, thus contributing to Type 2 error on those reinforcing outcome types in Experiment 1A. In Experiment 1B, we looked for systematic, outcome-related changes in the average luminance of the screen on genuine slot machines.

Method

To characterize this possible confounding role of luminance changes, we extracted screen-by-screen luminance values from the slot machine videos captured as part of Experiment 1A. We synchronized these luminance values to the event epochs defined in Experiment 1A. In performing these analyses, we were mindful that any luminance changes could be unique to the colors and associated luminance of the specific slot machine that we used, Buffalo Spirit (see Fig 3). For example, Buffalo Spirit displays a relatively light background for the large ‘reel spin’ area, and thus the symbols, including those associated with the bonus rounds, tend to be relatively dark. To address this possibility, we extracted screen luminance from archival data from a second authentic slot machine game in our laboratory, “Ice Empress” (Scientific Games Co., Las Vegas, NV). This game is similar to Buffalo Spirit in terms of its structural properties but employs a distinct visual aesthetic in which lighter symbols are presented on a darker screen background (see Fig 3).

Fig 3

Illustrations depicting the sequence of on-screen events on free-spin bonus features and winning outcomes, for the two slot machine models used in the present study.

These illustrations capture the approximate colour schemes and image dimensions of the key game elements. Researchers requiring actual screenshots should contact the corresponding author. Losses-disguised-as-wins follow similar in-game processes to the winning outcomes shown, and loss outcomes are similar to wins during the reel spin but do not contain any audiovisual feedback. Note the contrasting luminance properties for Buffalo Spirit (as a light screen background) and Ice Empress (with a darker screen background).

The luminance extraction for Ice Empress used a dataset from an unpublished behavioral experiment (Ferrari & Clark. [Unpublished]) in which 31 undergraduate participants played the slot machine for a 20-minute session. This study was conducted with ethical approval from the University’s Behavioral Research Ethics Board and written informed consent was taken from all participants. For clarity, we note the present analysis does not use any actual participant data from these sessions; we are solely using video capture of the slot machine’s monitor to examine game-level characteristics.

We modified the software used to produce the time series of slot machine outcomes in Experiment 1A, to convert each frame and pixel of the participants’ slot machine screen recordings into the CIELAB color space, separating three channels: L, the lightness (or “luminance”) scale, as well as a and b, the color components. For each frame, we extracted an average luminance (L) value for all pixels, ranging from 0 to 255. CIELAB is preferable to the RGB color space because it is perceptually uniform: changes in L correspond proportionally to changes in perceived screen brightness. Analyses were conducted using R (4.0.0).

We calculated these average L values as a function of the four outcome types in Experiment 1A, for all participants and all trials. For exploratory purposes, we plot these luminance levels for the full duration of the three phases of each ‘trial’ (gamble): the reel spin (i.e., prior to the outcome), the audiovisual feedback, and the subsequent inter-trial interval (Fig 4). Trial duration varied both between but also within outcome types, such that large error bars are evident towards the end of the three phases, due to the diminishing number of trials. We discarded frames exceeding that of the mean plus one standard deviation of trial duration, separately for each outcome by epoch. To facilitate a more direct comparison with Experiment 1A, we also plotted L values within the shorter epochs reflecting the Baseline, T1 and T2 phases (Fig 4); these epochs fall within the longer reel spin, audiovisual feedback, and inter-trial interval epochs, as depicted in Fig 1. The Baseline epochs were extracted from the reel spin phase prior to discarding frames, so that the Baseline would be immediately ‘contiguous’ with the T1 epoch; for this reason, the Baseline and reel spin plots do not perfectly align in Fig 4.

Fig 4

Buffalo spirit slot machine screen luminance by outcome phase.

Average screen luminance (L) component from CIELAB color space time-locked to event onset. L range = 0–255, Audiovisual Feedback Time ranged from 0–140,430 ms; a narrower range limited at 30,000 ms is presented for easier inspection and comparisons to T1. Ribbons represent one standard deviation above and below the mean of 7284 loss (79.2%), 909 win (9.9%), 942 loss-disguised-as-win (10.2%), and 62 bonus (0.7%) outcomes at a given point in Time. Baseline and Reel Spin outcomes do not align due to discarded frames in the latter.

Buffalo spirit

A number of interesting features are evident in the full time-courses depicted in Fig 4. First, during the reel spin, bonus rounds are associated with gradual decrease in screen luminance, which is not evident for losses, wins, and LDWs (see also Table 2). There are also notable differences in the durations of the reel spin (see Fig 4 and S1 Table), such that the spins leading up to the bonus rounds are substantially elongated due to the unique audiovisual accompaniment to each bonus symbol ‘landing’ [67]. The bonus symbols are roughly twice as large as the symbols that award regular wins, they are darker, and as they appear successively on three (or more) of the five reels (Fig 3), this likely accounts for the apparent incremental decrease in luminance.

Table 2

Averaged slot machine screen luminance, full epochs.

	N	Reel Spin	Audiovisual Feedback	Inter-trial Interval
Buffalo Spirit
Loss	7284	158.80 (4.65)	—	160.23 (4.87)
Win	909	159.94 (4.98)	156.15 (8.09)	158.04 (7.93)
LDW	942	159.38 (4.58)	149.27 (7.18)	160.43 (4.53)
Bonus	62	152.88 (3.92)	150.10 (9.49)	150.19 (2.90)
Ice Empress
Loss	3060	76.49 (6.95)	—	74.25 (7.06)
Win	406	78.01 (8.30)	85.96 (7.03)	81.64 (8.16)
LDW	292	76.69 (6.95)	77.21 (6.52)	79.05 (6.71)
Bonus	36	79.37 (5.73)	79.69 (9.87)	80.63 (5.37)

LDW = Loss-Disguised-as-a-Win.

Second, during the audiovisual feedback phase, a large dip in luminance was apparent on bonus rounds (about 20% of the full range of the luminance scale, or ≈50L out of 255). Bonus features are unique in that once triggered, the reels shrink, and a large dark blue border alerts the user to the free spins (Fig 3). This effect is the presumed source of the luminance dip. As the bonus spins begin, the reels return to full size, and there is a corresponding recovery in luminance to its previous level. Third, a luminance dip is also apparent following wins (about 10% of the full range of the luminance scale, or ≈25L out of 255), with a faster onset than the dip on bonuses. This likely represents the appearance of the flashing, colored borders around the winning paylines (Fig 3), which tend to be darker in color, given the lighter screen background of Buffalo Spirit. Lastly, during the inter-trial interval phase, the reels reset to the original winning combination of symbols prior to audiovisual onset (Fig 3). Thus, L at the inter-trial interval and baseline epochs are equivalent, and as such, the lower L on bonus rounds recurs during the inter-trial interval. As the machine is still during the inter-trial interval, the L values are stable throughout this phase for each outcome type (Fig 4).

The luminance fluctuations described above may impinge on the shorter epochs from Experiment 1A. As the Baseline epoch was defined as the final 200 ms of the reel-spin, the L at Baseline is lower on bonuses compared to the other three outcome types (Table 3), presumably due to the visual properties of the bonus symbols. During T1, the wins diverge from losses and losses-disguised-as-wins, drawing somewhat nearer to bonuses, which remained lower on average (Table 3). During T2, wins and losses-disguised-as-wins appeared to have a slightly higher average luminance than bonuses (Table 3).

Table 3

Buffalo spirit averaged slot machine screen luminance, measured epochs.

	N	Baseline	T1	T2
Loss	7284	160.50 (4.74)	160.42 (4.94)	—
Win	909	159.25 (4.76)	153.10 (10.20)	159.67 (5.52)
LDW	942	161.23 (4.51)	160.06 (5.44)	160.47 (4.74)
Bonus	62	148.27 (2.77)	149.43 (3.47)	151.75 (4.68)

LDW = Loss-Disguised-as-a-Win.

Ice empress

Ice Empress employs a darker background than Buffalo Spirit (Fig 3), and thus the overall screen luminance is considerably lower (≈75L) compared to Buffalo Spirit (≈150L). On Ice Empress, there is minimal evidence of luminance changes during the reel spin or the inter-trial interval phase (see Fig 5 and Table 2). However, during the audiovisual feedback phase, the bonus rounds were associated with a large spike in L (i.e., a rapid increase and recovery), which was also evident to a smaller extent following wins, and thus mirrored the luminance dips observed on Buffalo Spirit, as a game that uses a lighter theme. Ice Empress bonuses follow the analogous structure as Buffalo Spirit, but with the color scheme inverted such that the signaling border change is light relative to the reels (Fig 3), leading to a spike in luminance.

Fig 5

Ice Empress slot machine screen luminance by outcome phase.

Average screen luminance (L) component from CIELAB color space. L range = 0–255, Audiovisual Feedback Time ranged from 0–176,220 ms; a narrower range limited at 30,000 ms is presented for easier inspection. Ribbons represent one standard deviation above and below the mean of 3060 loss (82.0%), 406 win (10.9%), 229 loss-disguised-as-win (6.1%), and 36 bonus outcomes (1.0%) at a given point in Time.

Discussion

Slot machine outcomes were associated with systematic and multi-faceted changes in screen luminance, most notably as a sustained effect during the audiovisual feedback that accompanies bonus features and, to a lesser extent, winning outcomes. These fluctuations were relatively modest during the shorter windows (i.e., Baseline, T1, T2 epochs) from which pupil diameter was extracted in Experiment 1A, and within those shorter epochs, luminance was quite stable (Fig 4). Thus, although large-scale luminance fluctuations occurred during these two slot machine games, in our view there is not a clear ‘luminance-based’ explanation for the significant effects on pupil diameter that we observed in Experiment 1A in response to the bonus features (see General Discussion for further discussion).

One limitation pertaining to these analyses of screen luminance derived from video capture is that objective ‘on screen’ luminance is not the same as the brightness perceived by our study participants, which is affected by luminous flux incident on the eye, and subject to further variability as a function of eye movements [7]. The key question in Experiment 2 is whether the pronounced luminance changes characterized in the screen extraction analysis are sufficient to cause pupillary responses. This will clarify the role of luminance as a confound to reward-related pupillometry measures.

Experiment 2

Stimulus preparation

We created a stimulus set using scrambled screen shots from our two slot machines, taken at the moment of peak luminance fluctuations that were observed on wins (≈25L decrease on Buffalo Spirit, ≈25L increase on Ice Empress) and bonus rounds (≈50L decrease on Buffalo Spirit and ≈50L increase on Ice Empress). We created two further stimuli intended as manipulation checks, with a greater ≈100L luminance difference (decrease on Buffalo, increase on Ice Empress) to ensure the sensitivity of our pupillometry measurement. We presented these image stimuli to new participants during pupillary recording, in a laboratory study that was a hybrid of a visual psychophysics design but displayed on a slot machine monitor to equate the screen properties, seating distance, and ambient lighting with Experiment 1A.

Specifically, our stimulus set comprised six conditions: three luminance contrasts for each of the two slot machines (see Fig 6). The stimuli were constructed from screen shots of the two slot machine games, which were scrambled to ensure that the displayed events would not be perceived as exciting or financially rewarding. Luminance within these screenshots was manipulated using Adobe Photoshop (v.21.1.10). Each condition comprised a 5s baseline stimulus, henceforth termed Trial A, based on the screen prior to the key game event, and then a 5s stimulus, henceforth Trial B, that captured the peak luminance changes from the screen luminance analysis. The conditions were as follows: 1) Bonus Conditions, for which Trial B depicted a 50L decrease in the Buffalo Spirit, and a 50L increase in the Ice Empress condition; 2) Win conditions, for which Trial B depicted a 25L decrease in the Buffalo Spirit, and a 25L increase in the Ice Empress condition. Because the visual stimulation that accompanies wins varies in color, size, and patterns, the Trial B stimulus was created directly from stimulus A, using Photoshop. 3) The manipulation-check conditions, in which Trial A was identical to the win condition, and Trial B was the same as Trial A, but manipulated by ≈100L.

Fig 6

Scrambled images of slot machine outcomes.

L = objective luminance, cd/m2 = photometer measured brightness rating in situ ambient lighting.

We recruited 15 graduate students and university staff from the UBC Psychology Department (9 men and 6 women; mean age = 23.4, SD = 3.3), using the same exclusion criteria as in Experiment 1A. Additional inclusion criteria imposed by running the study during the COVID-19 pandemic was that participants could not be over 60 years of age, immunocompromised, and/or suffering from chronic diseases. This experiment used a different eye tracker than Experiment 1A, a decision that was enforced by SMI discontinuing product support. We used a mobile eye-tracker (Pupil Labs, Berlin, Germany), that is very similar to the SMI glasses with the exception that prescription lenses are not available. As such, participants with corrected-to-normal vision needed to wear contact lenses to avoid recording artifacts due to different pairs of glasses. Three participants were excluded due to equipment error. In one other participant, pupillary recording was lost after 16 of 30 blocks but the available data was valid and included. Thus, analyses are reported on 12 participants.

After giving informed consent, participants were seated in front of an authentic slot machine chassis wearing the eye tracking equipment. The sessions began with a 5-minute dark adaptation. The task was presented via PowerPoint (v.17.0) and fed through an HDMI cable to appear on the slot machine screen, a 22” LCD monitor (Wells-Gardner WGF2298A, McCook, IL). We employed a colorimeter (ColorCAL MKII, Cambridge Research Systems, Kent, UK) positioned at a viewing distance of 60cm from the slot machine screen to confirm the correspondence between our stimuli’s screen luminance (L) and the luminance of those stimuli emitted from the monitor (cd/m2) (see Fig 6). Additionally, the monitor emitted a maximum luminance rating of 129.25 cd/m2, and the in situ ambient lighting conditions of the laboratory was minimal (0.13 cd/m2). Other than the slot machine, the lighting conditions of the testing room were constant.

The six conditions were presented in blocks that comprised four repetitions of Trial A and Trial B, followed by a 30 s pupil recovery time in the dark after the offset of the fourth repetition (for comparable protocols, see [68, 69]). Each condition was presented 5 times, creating 30 blocks. To minimize eye movements, a small, white-dotted fixation point appeared in the center of the screen 0.5 s prior to the offset of a stimulus and disappeared with the onset of the next stimulus. Condition order was balanced using a Latin square design.

Pupil Lab’s Pupil Capture software extracted artefacts such as pupil detection (“confidence”) and pupil diameter from videos at 60 frames per second. High confidence values (max 1.0) indicated high quality of pupil detection for a given frame. Data points with confidence of less than .80 indicated potential blinks and saccades, and were discarded (5.93% of datapoints from Buffalo Spirit; 6.59% from Ice Empress). Epochs contained enough observations to calculate means, and thus missing data were not interpolated (i.e., missing data were treated as NaNs in the analysis).

Analyses were conducted separately for Buffalo Spirit and for Ice Empress using R (4.0.0). We calculated the mean pupil diameter in the last 200 ms of Trial A, and the mean pupil diameter in an interval from 200 ms to 2,500 ms on Trial B (see Fig 7), and our measure of pupil diameter change was the difference score from Trial B minus Trial A. Our defined time periods were in keeping with Experiment 1A and favorable to alternatives. For example, were we to calculate difference scores using two contiguous 2.5 s epochs, the latter 2.5 s of Trial A allows for gradual increases in pupil diameter. Specifically, pupils ‘escape’ from sustained light or ‘recover’ from darkness [5], and such events might occur, respectively, for Buffalo Spirit (where Trial A is lighter than Trial B) and Ice Empress (Trial A is darker than Trial B). Hence, the last 200 ms at Trial A occurring immediately prior to Trial B provides an appropriate time period of comparison. We aggregated the difference scores across the 5 (or fewer, for one participant) repetitions of each condition. We tested for trends across the sequence within each block, and these effects were non-significant and so we proceeded with the analysis on the aggregated difference scores.

Fig 7

Pupil time-course.

In the top row, lines indicate the mean pupil diameter from 200 to 2500 ms on Trial B for each participant (n = 12). In the bottom row, lines indicate the mean pupil diameter from 200 to 2500 ms on Trial B aggregated across participants. Pupil diameter is estimated as pixels as observed in the eye on the camera image, and is therefore not corrected for perspective.

Tests of univariate normality (Shapiro-Wilk) revealed residual non-normality (p < .05). Thus, as an initial test of whether difference score medians were significantly different from zero, we conducted the Wilcoxon signed-rank test, a non-parametric alternative to the sample t-test. Effect sizes were estimated using R as an alternative to Cohen’s d [70]. Mauchly’s test yielded non-significant results (p > .05), and thus no correction for non-sphericity was applied. Because ANOVA is generally robust to violations of normality [71], we did not correct for non-normality in our ANOVAs.

On Buffalo Spirit, there was a significant effect of luminance condition on the pupil diameter difference scores, F(2, 22) = 65.88, p < .001. Pairwise tests (Bonferroni correction to α = .017) found that the ≈100L manipulation check stimulus pair was associated with significantly greater pupil changes than the bonus (≈50L) and win (≈25L) stimulus pairs, which did not differ from one another. Testing each stimulus pair against a reference value of 0% change, the pupillary response was significant for each stimulus pair, with medium to large effect sizes (R2 = .28 to .76; Table 4).

Table 4

Difference score medians, IQRs, Wilcoxon signed-rank test p-values and effect sizes.

	Median	IQR	p	r	R 2
Buffalo Spirit
Bonus Condition	.78	.90	< .001	.55	.30
Win Condition	.66	1.34	< .001	.53	.28
Manipulation-check Condition	4.26	2.02	< .001	.87	.76
Ice Empress
Bonus Condition	-3.11	2.28	< .001	.87	.76
Win Condition	-1.87	1.97	< .001	.87	.76
Manipulation-check Condition	-4.40	1.94	< .001	.87	.76

N = 12.

The equivalent model for Ice Empress also found a significant effect of luminance conditions, F(2, 22) = 62.42, p < .001. Follow-up pairwise tests revealed that the manipulation condition produced significantly greater pupil changes than the bonus condition, and the win condition, and there was also a significantly greater response to the bonus condition compared to the win condition. The difference score medians per condition, representing pupil contraction (i.e., decreases in Trial B relative to Trial A), were each significantly less than the population median of zero, with large effects (R2 = .76; Table 4).

Although the pupillary changes were clearly of greatest magnitude in the manipulation-check condition, the two stimulus pairs that were based on realistic in-game luminance fluctuations were sufficient to trigger changes in pupil diameter. This highlights a need to account for incidental luminance changes in future studies employing ecologically-valid gambling products.

General discussion

We investigated physiological correlates of genuine slot machine gambling using pupillary measures obtained from mobile eye tracking glasses. In Experiment 1A, we found significant pupil dilation (a marker of sympathetic arousal) immediately after participants received free spin bonus feature outcomes. In Experiment 1B, we noted differences in luminance between slot machine outcomes as a potential confound to our arousal-related effects. In Experiment 2, luminance fluctuations using scrambled slot machine displays and scaled on the actual game-related changes were sufficient to trigger reliable pupil responses.

Do pupillary responses to bonus rounds reflect sympathetic arousal or a reflexive response to lighting changes?

We observed significant pupil dilation in response to receiving free spin bonus features in Experiment 1A. There was no evidence that pupil diameter was sensitive to reinforcement from wins and losses-disguised-as-wins (relative to losses). The rarer, more-salient bonus features may produce a greater sense of excitement than conventional winning outcomes (and losses-disguised-as-wins), to which gamblers may rapidly habituate. By this account, modern slot machines might not be predominantly physiologically arousing; rather, their appeal could be based on the pursuit of occasional events such as bonus rounds or larger jackpot wins, or on providing interesting or immersive gameplay features.

The significant increase in pupil diameter for bonus features at T1 relative to baseline did not evidently co-occur with any decrease in screen luminance for bonuses at T1. Moreover, luminance values for the different slot machine outcomes varied little from baseline to T1 to T2. Thus, although we found marked changes in luminance during the reinforcing outcomes (discussed in the next section), such changes occurred outside the time course of our measured epochs in Experiment 1A. This favors an arousal- or excitement-based account of the pupillary effect to bonus features, rather than a reflexive response to incidental changes in screen brightness.

Nevertheless, pupil diameter can be influenced by several factors. For example, the duration of the reel spin was substantially longer during free spin bonus features than other outcomes (see Fig 4 and S1 Table). Because pupil diameter is sensitive to uncertainty and anticipation of reward [72, 73], it is unclear from our findings whether the effects of bonus features are due to reward per se, or due to the anticipation of reward. We also note that the symbols that trigger bonus rounds on ‘Buffalo Spirit’ are physically larger and accompanied by loud and distinctive auditory flourishes. Recent research has shown that pupil dilation can be elicited by auditory and/or visual stimuli, and further modulated by stimulus salience [74]. Thus, the observed effects of bonus rounds on pupil size may be due to their auditory and visual features, as well as their relative salience and infrequency. In contrast to bonuses, loss outcomes occurred on around 80% of trials, and this relative predictability is less than ideal for using the losses as the reference category for all three types of reinforcing outcomes. The low frequency of bonus rounds also means that these events occurred in only a subset of sessions and participants. Because we used a genuine slot machine, we could not control these outcome frequencies, and our analyses involving bonus features draw upon a very small number of observations (see S1 Table). This likely diminished the reliability of the reported coefficient estimates.

What game events trigger luminance fluctuations in authentic slot machines?

We observed luminance fluctuations during the delivery of wins and bonuses, in Buffalo Spirit (decreasing luminance; Fig 4) and Ice Empress (increasing luminance; Fig 5). In line with the pupil light reflex [5], the onset of darker stimuli on a slot machine monitor increased pupil diameter (Buffalo Spirit; Table 4) and the onset of brighter stimuli decreased pupil diameter (Ice Empress; Table 4). These data suggest that meaningful on-screen luminance changes accompany wins and bonuses, and therefore that caution is required in attributing the pupillary effects on positively-reinforcing outcomes to the sympathetic nervous system. Nonetheless, as the luminance fluctuations on bonuses occurred after T1, it is unlikely that the pupil dilation observed at the onset of the bonus audiovisuals (T1) was related to changes in screen luminance.

The strength of the pupil response scales with the magnitude of luminance change: the manipulation-check condition that offered the greatest luminance difference (≈100L) produced significantly greater pupil response than the two stimulus pairs derived from the slot machine game, at ≈25L for wins and ≈50L for bonuses. Further, the pupil response appears to be more sensitive to slot machine screen brightness than darkness. Specifically, the bonus condition produced stronger effects than the win condition in Ice Empress (but not in Buffalo Spirit). This is consistent with research suggesting that pupils constrict more immediately to brightness and recover more gradually in darkness [5]. It is likely that slot machine manufacturers use luminance to increase the saliency of positively-reinforcing outcomes. The practical implication of our work is that future slot machine research utilizing pupillometry should consider luminance fluctuations in the context of the specific gambling product being used, and account for luminance in terms of both magnitude and direction of change.

There is increasing interest in the use of pupillometry and other eye tracking metrics in relation to gambling [75]. Refining techniques that can separate the pupil effects of cognitive activity vs. light input in realistic settings (e.g., Marshall’s Index of Cognitive Activity; [76-78]) would further enhance the use of pupillometry for indexing surprise, excitement or other emotional and motivational processes that are elicited during slot machine gambling [79]. This would add to the growing body of work that examines physiological states and how individual differences in psychophysiological reactivity relate to risk of gambling harm.

Limitations

Experiment 1A sought to examine slot machine gambling in an ecologically-valid manner. We could not control any perceptual or phenomenal differences in slot machine events or areas of the game screen, and we did not correct for individual differences in pupillary foreshortening [80]. The hardware differences between the two studies may have differentially impacted foreshortening error (see S1 Text). On bonus features, unique audiovisual features occurred during the baseline pupil measurement, such that this epoch is unlikely to represent a neutral baseline. In contrast, Experiment 2 was more tightly controlled, and the stimulus pairs were chosen to capture the peak luminance effects, but those observations may not generalize to real-life slot machine gambling, which contains other confounding factors besides luminance (e.g., reinforcement schedules, slot machine sounds, ambiance of the casino). Several additional limitations are discussed in the S1 Text.

Conclusion

Mobile eye tracking technology offers a valuable tool for exploring gamblers’ reactions to gambling products and environments. Our results suggest that reinforcement from slot machine bonus rounds is associated with pupillary dilation, but that future studies should pay careful attention to fluctuations in luminance as a possible confound in the interpretation of pupillary effects. Researchers should expect that these fluctuations may be game-specific, and their assessment will require both high resolution video capture and event-related analyses.

(DOCX)

Click here for additional data file.

Outcome frequencies and durations.

Loss outcomes do not involve any audiovisual feedback and thus we could not specify a T2 event (that was distinct from T1). Bonuses include all free spins during the Audiovisuals phase. LDW = Loss-Disguised-as-a-Win.

(DOCX)

Click here for additional data file.

Correlation between corrected and uncorrected pupil diameter.

Within-subjects correlations for each of the 53 participants in Experiment 1 are shown. Data labels depict the maximum, median, and minimum correlations.

(TIF)

Click here for additional data file.

28 Feb 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Manuel Spitschan

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Competing Interests section:

(I have read the journal's policy and the authors of this manuscript have the following competing interests:

The Centre for Gambling Research at UBC receives funding from the Province of British Columbia and the British Columbia Lottery Corporation (BCLC), a Canadian Crown Corporation. The slot machines used in the present study were provided to the Centre for Gambling Research by the BCLC.

EHLO has received a speaker honorarium from the Massachusetts Council on Compulsive Gambling (USA). She has accepted travel or accommodation for speaking engagements from the National Council for Responsible Gambling (USA), the International Multidisciplinary Symposium on Gambling Addiction (Switzerland), and the Responsible Gambling Council (Canada). She has not received any further direct or indirect payments from the gambling industry or groups substantially funded by gambling.

MAF has received a speaker honorarium from the British Columbia Lottery Corporation (BCLC).

MVC has received a speaker honorarium from the Responsible Gaming Association of New Mexico (USA).

LC is the Director of the Centre for Gambling Research at UBC. LC has received speaker/travel honoraria from the National Association of Gambling Studies (Australia) and the International Center for Responsible Gaming (USA). He has received academic consulting fees from Gambling Research Exchange Ontario (Canada), GambleAware (UK), and the International Center for Responsible Gaming (USA). He has not received any further direct or indirect payments from the gambling industry or groups substantially funded by gambling. He has received royalties from Cambridge Cognition Ltd. relating to neurocognitive testing.

AJK, WSM, KIM, JF and MS report no conflicts of interest.)

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

3. We note that you have referenced) (Ferrari & Clark, unpublished) which has currently not yet been accepted for publication. Please remove this from your References and amend this to state in the body of your manuscript: (ie “Bewick et al. [Unpublished]”) as detailed online in our guide for authors

http://journals.plos.org/plosone/s/submission-guidelines#loc-reference-style

4. We note that Figure 3 in your submission contain copyrighted images. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

a. You may seek permission from the original copyright holder of Figure 3 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

5. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Do pupillary responses during authentic slot machine use reflect arousal or screen luminance fluctuations? A proof-of-concept study

===================================================================================================================================

The manuscript explores whether pupillometry can give insight into arousal in a slot machine gambling scenario and asks whether arousal-linked pupil responses can be differentiated from those caused by intense audiovisual feedback. The task is ecologically valid--real slot machines were used--and participants (n = 53) were themselves slot machine gamblers. In the first experiment, pupil data related to wins, losses-disguised-as-wins, and bonus feature events were compared to a loss baseline. The pupil response was greatest for bonus feature events, which were characterised by intense audiovisual feedback. In the second part of the first experiment, real-time luminance data are extracted from the displays. Analyses and visualisations of these data confirmed that there are systematic effects of luminance between the different events analysed in the first part of the experiment. Finally, experiment 2 used scrambled stills of the slot machine events to confirm that pupil responses were indeed triggered by luminance information.

The writing is clear, easy to follow, and close to publication standard. I have the following comments / suggestions:

Comments / suggestions

----------------------

1. It is interesting to see cognitive pupillometry applied in the context of a slot machine gambling task. I feel that the narrative could be enhanced if the authors were able to place their findings within the wider context of studies that attempt to use pupillometry in realistic settings. For example, what do the author's think of Marshall's (2008) index of cognitive activity (which purports to factor out luminance changes), and Bhavsar et al.'s control room operator study?

Marshall, S. P. (2002). The Index of Cognitive Activity: Measuring cognitive workload. Proceedings of the IEEE 7th Conference on Human Factors and Power Plants, 7-5-7–9. https://doi.org/10.1109/HFPP.2002.1042860

Bhavsar, P., Srinivasan, B., & Srinivasan, R. (2016). Pupillometry based real-time monitoring of operator’s cognitive workload to prevent human error during abnormal situations. Industrial and Engineering Chemistry Research, 55(12), 3372–3382. https://doi.org/10.1021/acs.iecr.5b03685

2. In Experiment 2 (p20), the author's state 'Data points with confidence of less than .80 indicated potential blinks and saccades, and were discarded (5.93% of datapoints from Buffalo Spirit; 6.59% from Ice Empress). Epochs contained enough observations to calculate means, and thus missing data were not interpolated.' -- were the discarded data points treated as NaNs? It would be interesting to see the time-courses in addition to the summary statistics presented in Table 4.

3. The main conclusion of the discussion (p25) states, 'The practical implication of our work is that future slot machine research utilizing pupillometry should account for luminance in terms of magnitude and direction of change.'. I feel that this statement is rather obvious, and that it applies more generally to any research where cognitive pupillometry is used in real life scenarios (see first comment). I therefore think the authors should comment more on why they think this is important -- what exactly is the promise, here? If there was a perfect system that could disentangle luminance- and arousal-related pupil responses, what more could we learn about slot machine gambling behaviour?

4. The author's admit that they did not correct for pupil foreshortening. Was the risk of this different for the two eye trackers that were used? I know that the Pupil Core system uses a 3-D model of pupil size that is robust to the effects of gaze position at the population level (Petersch & Dierkes, 2021), but I'm not sure about the SMI system. Some more info here would be useful. Also, as no corrections were applied, and as the overall findings are not particularly striking, I think the manuscript would benefit from further analyses to explore whether pupil size was systematically related to gaze position.

Petersch, B., & Dierkes, K. (2021). Gaze-angle dependency of pupil-size measurements in head-mounted eye tracking. Behavior Research Methods. https://doi.org/10.3758/s13428-021-01657-8

Brisson, J., Mainville, M., Mailloux, D., Beaulieu, C., Serres, J., & Sirois, S. (2013). Pupil diameter measurement errors as a function of gaze direction in corneal reflection eyetrackers. Behavior Research Methods. https://doi.org/10.3758/s13428-013-0327-0

Martin, J. T., Whittaker, A. H., & Johnston, S. J. (2020). Component processes in free-viewing visual search: Insights from fixation-aligned pupillary response averaging. Journal of Vision, 20(7), 5. https://doi.org/10.1167/jov.20.7.5

Reviewer #2: The study examines pupillary responses during authentic slot machine. The results show some differences in pupil size in some comparisons, and these observed differences in pupil size cannot be simply explained by the differences of overall luminance level. Overall, I think this study is interesting and the motivation is clear, particularly it is important to study in the real-world gambling situation. However, there are a range of factors that also influence pupil size, and differences in visual, auditory and temporal features between different conditions may have confounded the observed results.

My major concern comes with the feedback differences in visual and audio features between 4 conditions. If I understand correctly, the critical time epoch should be the differences in pupil size between T1 and T2 (feedback epoch), as any results after T2 could overlap with the next run temporally. However, feedback durations are different among different conditions, with longest for wins, shortest for LDWs, no feedback for losses. First, it is unclear the duration “range” of the feedback in these 4 conditions, also it is not clear about bonus feedback. So, it is better to clearly describe this information in Table (duration range). Furthermore, the main comparison is between the win/LDW/bonus and loss conditions. If there is no audiovisual feedback in the loss condition, the T1-T2 duration for the loss condition is likely overlapping with the next run temporally, how to perform this comparison properly? Thus, if the win and loss (or LDW) conditions have a relatively similar trial features (e.g., duration), it may be better to compare these two conditions only.

The authors conducted some other experiments to control luminance influence on pupil size, which is appropriate. However, there are many other visual factors that also influence pupil size (e.g., Barbur, J. L. (2004). Learning from the pupil-studies of basic mechanisms and clinical applications. The visual neurosciences, 1, 641-656). Besides, pupil dilation is observed after acoustic sounds, and the magnitude of dilation is scaled with stimulus intensity (also arousing level). All these factors need to be taken into consideration to make sure that the observed differences in pupil size are not simply mediated by these differences.

Also, trial number also radically different among different conditions, as most trials probably are loss trials (~60-70 %). Does trial number could affect the observed effects in pupil size?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Joel T. Martin

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

19 May 2022

Date: 2022-May-06

Ms. No.: PONE-D-22-00740

Academic Editor: Dr. Manuel Spitschan

Journal: PLOS ONE

Title: Do pupillary responses during authentic slot machine use reflect arousal or screen luminance fluctuations? A proof-of-concept study

We would like to thank the Academic Editor and Reviewers for their helpful feedback. We have worked to revise our submission in light of the comments and suggestions provided, and we believe the revised manuscript has benefitted substantially from the feedback.

Response to Reviewer 1’s Comments

1. It is interesting to see cognitive pupillometry applied in the context of a slot machine gambling task. I feel that the narrative could be enhanced if the authors were able to place their findings within the wider context of studies that attempt to use pupillometry in realistic settings. For example, what do the author's think of Marshall's (2008) index of cognitive activity (which purports to factor out luminance changes), and Bhavsar et al.'s control room operator study?

RESPONSE: These are good suggestions. We have added the following text in the Discussion to enhance our narrative (the text below also addresses the reviewer’s comment #3).

“There is increasing interest in the use of pupillometry and other eye tracking metrics in relation to gambling (Li et al., 2021). Refining techniques that can separate the pupil effects of cognitive activity vs. light input in realistic settings (e.g., Marshall’s Index of Cognitive Activity; Bhavsar et al., 2016; Marshall, 2002; Vogels et al., 2018) would further enhance the use of pupillometry for indexing surprise, excitement or other emotional and motivational processes that are elicited during slot machine gambling (Preuschoff, Hart, & Einhäuser, 2011). This would add to the growing body of work that examines physiological states and how individual differences in psychophysiological reactivity relate to risk of gambling harm.” (Page 25-26)

2. In Experiment 2 (p20), the author's state 'Data points with confidence of less than .80 indicated potential blinks and saccades, and were discarded (5.93% of datapoints from Buffalo Spirit; 6.59% from Ice Empress). Epochs contained enough observations to calculate means, and thus missing data were not interpolated.' -- were the discarded data points treated as NaNs? It would be interesting to see the time-courses in addition to the summary statistics presented in Table 4.

RESPONSE: We apologize for any confusion. To clarify, the discarded data points were treated as NaNs. We have revised the text to state:

“Epochs contained enough observations to calculate means, and thus missing data were not interpolated (i.e., missing data were treated as NaNs in the analysis).” (Page 20)

As for including pupil time-courses, we thank the reviewer for this excellent suggestion. We now include two additional figures (Fig 7) representing pupil time-courses in our manuscript (see below).

Fig 7. Pupil time-course. In the top row, lines indicate the mean pupil diameter from 200 to 2500 ms on Trial B for each participant (n = 12). In the bottom row, lines indicate the mean pupil diameter from 200 to 2500 ms on Trial B aggregated across participants. Pupil diameter is estimated as pixels as observed in the eye on the camera image, and is therefore not corrected for perspective.

3. The main conclusion of the discussion (p25) states, 'The practical implication of our work is that future slot machine research utilizing pupillometry should account for luminance in terms of magnitude and direction of change.'. I feel that this statement is rather obvious, and that it applies more generally to any research where cognitive pupillometry is used in real life scenarios (see first comment). I therefore think the authors should comment more on why they think this is important -- what exactly is the promise, here? If there was a perfect system that could disentangle luminance- and arousal-related pupil responses, what more could we learn about slot machine gambling behaviour?

RESPONSE: We agree with the referee, but in our view the impact of luminance fluctuations, and the importance of monitoring and adjusting for these fluctuations, is unlikely to be obvious to the field of gambling studies (e.g. https://www.sciencedirect.com/science/article/pii/S0167876019305434 and preprint http://europepmc.org/article/PPR/PPR478750 ). There is increasing interest in the use of pupillometry and other eye tracking metrics in relation to gambling, but modern gambling products sit in a grey zone between traditional cognitive tasks and modern research conducted in naturalistic settings: slot machines (and other gambling products) represent programmed environments that contain a specified and discrete set of events (wins, losses etc) but these are commercial products where academic researchers typically have minimal access to the underlying code, and so are forced to use ‘naturalistic’ designs. As recommended, we added a section on the impact of our work in the Discussion (Page 25-26):

“There is increasing interest in the use of pupillometry and other eye tracking metrics in relation to gambling (Li et al., 2021). Refining techniques that can separate the pupil effects of cognitive activity vs. light input in realistic settings (e.g., Marshall’s Index of Cognitive Activity; Bhavsar et al., 2016; Marshall, 2002; Vogels et al., 2018) would further enhance the use of pupillometry for indexing surprise, excitement or other emotional and motivational processes that are elicited during slot machine gambling (Preuschoff, Hart, & Einhäuser, 2011). This would add to the growing body of work that examines physiological states and how individual differences in psychophysiological reactivity relate to risk of gambling harm.”

4. The author's admit that they did not correct for pupil foreshortening. Was the risk of this different for the two eye trackers that were used? I know that the Pupil Core system uses a 3-D model of pupil size that is robust to the effects of gaze position at the population level (Petersch & Dierkes, 2021), but I'm not sure about the SMI system. Some more info here would be useful. Also, as no corrections were applied, and as the overall findings are not particularly striking, I think the manuscript would benefit from further analyses to explore whether pupil size was systematically related to gaze position.

RESPONSE: Thank you. These are valid concerns, and we have thought carefully about how to address this issue in light of the points raised. We attempted the recommended foreshortening correction procedure and generated the results reported in S1 Figure below. However, these efforts revealed a critical problem in the underlying data. Namely, we do not have access to the original raw data for point-of-regard analyses. During the analysis of Study 1, SMI’s take-over by Apple caused them to discontinue academic support, and the output files that we held at that time only included point-of-regard data that had been manually mapped using SMI’s proprietary software package, BeGaze. (This was also the impetus for our switch to a PupilLabs eye tracker for Study 2). As a result, the small proportion of fixations (<1%) that were directed outside of the slot machine screen were mapped to the nearest point on the outer edge of the reference image rather than the location where they truly occurred. As such, our foreshortening correction cannot properly account for fixations occurring away from the slot machine screen, where the largest foreshortening errors would occur.

We state our procedures and explain why we ultimately elected not to correct for pupil foreshortening in S1 Text:

“The Pupil Core system employed in Experiment 2 uses a three dimensional model of pupil size that is more robust against foreshortening error (Petersch & Dierkes, 2021). Experiment 1 used SMI’s BeGaze analysis software, which provided less robust estimates of pupil diameter. Although participants in that study spent the overwhelming majority of the task staring straight ahead at the slot machine screen (Murch et al., 2020), results in Experiment 1 could have been impacted by participants’ gaze positions during a particular event relative to the position of each eye tracking camera (i.e., foreshortening errors).

Using available point-of-regard data from Experiment 1, we examined the correlation between uncorrected pupil diameter measurements (in millimeters), and measurements which had been corrected for pupillary foreshortening using a validated linear regression procedure (Brisson et al., 2013; Martin et al., 2020). The individual correlations between corrected and uncorrected pupil diameters are presented in S1 Fig. In general, there was a high correlation between corrected and uncorrected pupil diameters. In the median participant, corrected pupil diameter explained a majority (r2 = 0.73) of the variation in uncorrected pupil diameter, indicating that approximately 27% of variance in pupil diameter was explained by the foreshortening correction model. For comparison, in Brisson et al (2013), 20% of the variance in uncorrected pupil diameter measurements could be explained by the foreshortening-correction regression model.

Crucially, the abrupt closure of SMI in 2017 prevented us from accessing the raw point-of-regard data (prior to our mapping of the data using SMI’s proprietary Semantic Gaze Mapping software), and this impaired the potential utility of our ‘corrected’ pupil diameter measurements. For these data, Semantic Gaze Mapping entailed manually identifying each point-of-regard using a reference image. The borders on this reference image are limited to the size of the slot machine screen, meaning that data pertaining to the visual periphery were lost. As such, our correction models lacked the necessary point-of-regard data needed to fix the most serious instances of pupillary foreshortening. These regression-based foreshortening corrections thus created potentially-misleading measurements for our data. Our analyses are therefore reported without correcting for pupillary foreshortening.”

S1 Fig. Correlation between corrected and uncorrected pupil diameter. Within-subjects correlations for each of the 53 participants in Experiment 1 are shown. Data labels depict the maximum, median, and minimum correlations.

The reviewer also raises a good point that the risk of foreshortening bias differs between the two experiments (based on the different eye tracking hardware), and we refer to S1 Text in our Limitations (page 26):

“We could not control any perceptual or phenomenal differences in slot machine events or areas of the game screen, and we did not correct for individual differences in pupillary foreshortening [80]. The hardware differences between the two studies may have differentially impacted foreshortening error (see S1 Text).”

Response to Reviewer 2’s Comments

1. My major concern comes with the feedback differences in visual and audio features between 4 conditions. If I understand correctly, the critical time epoch should be the differences in pupil size between T1 and T2 (feedback epoch), as any results after T2 could overlap with the next run temporally. However, feedback durations are different among different conditions, with longest for wins, shortest for LDWs, no feedback for losses. First, it is unclear the duration “range” of the feedback in these 4 conditions, also it is not clear about bonus feedback. So, it is better to clearly describe this information in Table (duration range). Furthermore, the main comparison is between the win/LDW/bonus and loss conditions. If there is no audiovisual feedback in the loss condition, the T1-T2 duration for the loss condition is likely overlapping with the next run temporally, how to perform this comparison properly? Thus, if the win and loss (or LDW) conditions have a relatively similar trial features (e.g., duration), it may be better to compare these two conditions only.

RESPONSE: We thank the reviewer for their careful consideration of this analysis. We agree that the different degrees of overlap between epochs referencing the start (T1) and end (T2) of audiovisual feedback limit our ability to discern event-related changes in pupil diameter. This is particularly the case for losses-disguised-as-wins. We have included the ranges for feedback duration in S1 Table under the column “Duration of Audiovisuals”. We have also indicated in the manuscript that, since loss trials do not have audiovisual feedback, there is no T2 event for this trial type present in the analyses. The fixed-effects regression approach we employed was partly chosen for its suitability to cases where some outcome types may not occur, or may not occur for all participants (Allison, 2012).

2. The authors conducted some other experiments to control luminance influence on pupil size, which is appropriate. However, there are many other visual factors that also influence pupil size (e.g., Barbur, J. L. (2004). Learning from the pupil-studies of basic mechanisms and clinical applications. The visual neurosciences, 1, 641-656). Besides, pupil dilation is observed after acoustic sounds, and the magnitude of dilation is scaled with stimulus intensity (also arousing level). All these factors need to be taken into consideration to make sure that the observed differences in pupil size are not simply mediated by these differences.

RESPONSE: Thank you for these excellent points. The reviewer is correct that pupil diameter may be impacted by other factors besides luminance and slot machine outcomes. Unfortunately, we do not have access to factors such as acoustic data in these 2 studies. Disentangling these factors is beyond the scope of the present study and would be best reserved for future directions. As such, we acknowledge the reviewer’s concerns as limitations in the paper (page 24):

“Nevertheless, pupil diameter can be influenced by several factors. For example, the duration of the reel spin was substantially longer during free spin bonus features than other outcomes (see Fig 4 and S1 Table). Because pupil diameter is sensitive to uncertainty and anticipation of reward (Nassar et al., 2012; Pietrock et al., 2019), it is unclear from our findings whether the effects of bonus features are due to reward per se, or due to the anticipation of reward. We also note that the symbols that trigger bonus rounds on ‘Buffalo Spirit’ are physically larger and accompanied by loud and distinctive auditory flourishes. Recent research has shown that pupil dilation can be elicited by auditory and/or visual stimuli, and further modulated by stimulus salience (Wang & Munoz, 2015). Thus, the observed effects of bonus rounds on pupil size may be due to their auditory and visual features, as well as their relative salience and infrequency.”

3. Also, trial number also radically different among different conditions, as most trials probably are loss trials (~60-70 %). Does trial number could affect the observed effects in pupil size?

RESPONSE: This is a good observation. As shown in S1 Table under the column “Total Events”, the number of trials does vary substantially across outcome types: Loss (n = 7451), Win (n = 876), losses-disguised-as-wins (n = 1016), and Bonus (n = 58). This discrepancy is because we cannot control the frequency of different outcomes and an unavoidable limitation of using a genuine slot machine in the experiment. We acknowledge this limitation and the potential consequences of infrequent bonuses and frequent losses:

“In contrast to the bonuses, loss outcomes occurred on around 80% of trials, and this relative predictability is less than ideal for using the losses as the reference category for all three types of reinforcing outcomes. The low frequency of bonus rounds also means that these events occurred in only a subset of sessions and participants. Because we used a genuine slot machine, we could not control these outcome frequencies, and our analyses involving bonus features draw upon a very small number of observations (see S1 Table). This likely diminished the reliability of the reported coefficient estimates.” (Page 24)

Response to the Academic Editor

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

RESPONSE: Thank you for pointing this out. Efforts have been made to ensure that the manuscript is meeting PLOS ONE’s style requirements.

2. Thank you for stating the following in the Competing Interests section:

(I have read the journal's policy and the authors of this manuscript have the following competing interests:

The Centre for Gambling Research at UBC receives funding from the Province of British Columbia and the British Columbia Lottery Corporation (BCLC), a Canadian Crown Corporation. The slot machines used in the present study were provided to the Centre for Gambling Research by the BCLC.

EHLO has received a speaker honorarium from the Massachusetts Council on Compulsive Gambling (USA). She has accepted travel or accommodation for speaking engagements from the National Council for Responsible Gambling (USA), the International Multidisciplinary Symposium on Gambling Addiction (Switzerland), and the Responsible Gambling Council (Canada). She has not received any further direct or indirect payments from the gambling industry or groups substantially funded by gambling.

MAF has received a speaker honorarium from the British Columbia Lottery Corporation (BCLC).

MVC has received a speaker honorarium from the Responsible Gaming Association of New Mexico (USA).

LC is the Director of the Centre for Gambling Research at UBC. LC has received speaker/travel honoraria from the National Association of Gambling Studies (Australia) and the International Center for Responsible Gaming (USA). He has received academic consulting fees from Gambling Research Exchange Ontario (Canada), GambleAware (UK), and the International Center for Responsible Gaming (USA). He has not received any further direct or indirect payments from the gambling industry or groups substantially funded by gambling. He has received royalties from Cambridge Cognition Ltd. relating to neurocognitive testing.

AJK, WSM, KIM, JF and MS report no conflicts of interest.)

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests). If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

RESPONSE: As requested, we have included the following statement in our Competing Interests section: “This does not alter our adherence to PLOS ONE policies on sharing data and materials” (see Cover Letter).

3. We note that you have referenced) (Ferrari & Clark, unpublished) which has currently not yet been accepted for publication. Please remove this from your References and amend this to state in the body of your manuscript: (ie “Bewick et al. [Unpublished]”) as detailed online in our guide for authors

http://journals.plos.org/plosone/s/submission-guidelines#loc-reference-style

RESPONSE: As suggested, (Ferrari & Clark. [Unpublished]) is in the body of our manuscript and excluded from the References.

4. We note that Figure 3 in your submission contain copyrighted images. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

RESPONSE: The copyright holders (WMS / Scientific Games) declined our copyright request. As such, we have supplied a replacement figure that we created (see below, Fig 3). We confirm that Fig 3 is entirely created by the authors and does not contain any copyrighted material, illustrations, designs, etc.

Fig 3. Illustrations depicting the sequence of on-screen events on free-spin bonus features and winning outcomes, for the two slot machine models used in the present study. These illustrations capture the approximate colour schemes and image dimensions of the key game elements. Researchers requiring actual screenshots should contact the corresponding author. Losses-disguised-as-wins follow similar in-game processes to the winning outcomes shown, and loss outcomes are similar to wins during the reel spin but do not contain any audiovisual feedback. Note the contrasting luminance properties for Buffalo Spirit (as a light screen background) and Ice Empress (with a darker screen background).

5. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

RESPONSE: Thank you. Captions are now included for the Supporting Information files at the end of our manuscript. In-text citations are updated accordingly.

Submitted filename: reviewer response letter.docx

Click here for additional data file.

13 Jul 2022

Do pupillary responses during authentic slot machine use reflect arousal or screen luminance fluctuations? A proof-of-concept study

PONE-D-22-00740R1

Dear Dr. Kim,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Manuel Spitschan

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: I appreaciate the work that has been done to address the issues raised, in particular the supplementary info on pupil forshortening corrections. I have no furter comments or suggestions.

Reviewer #2: The authors have addressed all my previous concerns, but as mentioned, many factors known to change pupil size are different among different conditions (e.g., visual and auditory attributes, trial number, trial duration). I'm less certain about their conclusion, nevertheless, it is indeed needed to see more pupil results in the real world data in gambling.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Joel T. Martin

Reviewer #2: No

**********

15 Jul 2022

PONE-D-22-00740R1

Do pupillary responses during authentic slot machine use reflect arousal or screen luminance fluctuations? A proof-of-concept study

Dear Dr. Kim:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Manuel Spitschan

Academic Editor

PLOS ONE