The role of perceptual processing in the oddball effect revealed by the Thatcher illusion

When a novel stimulus (oddball) appears after repeated presentation of an identical stimulus, the oddball is perceived to last longer than the repeated stimuli, a phenomenon known as the oddball effect. We investigated whether the perceptual or physical differences between the repeated and oddball stimuli are more important for the oddball effect. To manipulate the perceptual difference while keeping their physical visual features constant, we used the Thatcher illusion, in which an inversion of a face hinders recognition of distortion in its facial features. We found that the Thatcherized face presented after repeated presentation of an intact face induced a stronger oddball effect when the faces were upright than when they were inverted (Experiment 1). However, the difference in the oddball effect between face orientations was not observed when the intact face was presented as the oddball after repeated presentation of a Thatcherized face (Experiment 2). These results were replicated when participants performed both the intact-repeated and Thatcherized-repeated conditions in a single experiment (Experiment 3). Two control experiments confirmed that the repeated presentation of the preceding stimuli is necessary for the difference in duration distortion to occur (Experiments 4 and 5). The results suggest the considerable role of perceptual processing in the oddball effect. We discuss the discrepancy in the results between the intact-repeated and Thatcherized-repeated conditions in terms of predictive coding.

A prevailing theory posits that the oddball effect is primarily caused by the duration distortion of the repeated rather than the oddball stimuli.This idea is based on the findings that perceived duration can be reduced by the repetition of a stimulus.For example, the first stimulus in a sequence of repeated stimuli is perceived to last longer than the subsequent ones (Cai et al., 2015;Pariyadath & Eagleman, 2007;Rose & Summers, 1995).Moreover, the duration of the second stimulus is perceived as shorter when identical stimuli are repeated than when two different stimuli are presented (Birngruber, Schröter, & Ulrich, 2015a;Matthews, 2011;Noguchi & Kakigi, 2006).Repetition of identical stimuli is known to reduce the neural response in the sensory cortex, which is known as repetition suppression (Henson & Rugg, 2003).Given this parallel relationship between perceived duration and neural response, studies have suggested that a repeated presentation of the standard stimuli reduces the perceived duration through repetition suppression, which leads to the apparent duration expansion of the oddball stimulus (Eagleman & Pariyadath, 2009;Pariyadath & Eagleman, 2007).
It has been suggested that predictive coding, which posits that the brain generates a prediction about forthcoming stimulus based on prior sensory events and updates the prediction based on the discrepancy between the predicted and actual inputs to minimise the prediction error at each processing stage (Friston, 2005;Rao & Ballard, 1999), plays an important role in the repetition suppression account of the oddball effect.Supporting this, the size of the oddball effect has been shown to increase as the visual feature (i.e., orientation) of the oddball stimulus differs more from the standard stimuli (Pariyadath & Eagleman, 2012;Schindel et al., 2011) and as the number of repeated stimuli before the oddball stimulus increases (Pariyadath & Eagleman, 2012).Furthermore, Schindel et al. (2011) reported that the difference between repetitive and oddball stimuli, rather than the apparent intensity of oddball stimuli, explained the magnitude of the oddball effect more.These results imply that the oddball effect reflects the difference between the predicted and actual inputs through predictive coding processes (Pariyadath & Eagleman, 2012;Schindel et al., 2011).
While previous studies have shown that violation of repetition prediction plays an important role, sufficient prediction error for inducing the oddball effect remains unclear.In particular, it is not clear whether the perceptual or physical differences between the repeated and the oddball stimuli are important for the oddball effect to occur.To address this question in the present study, we used the Thatcher illusion, in which an inversion of a face hinders recognition of distortion in its facial features (e.g., Thompson, 1980).Fig. 1 shows examples of the face stimulus.Although there is no difference in the image properties between the upright and the inverted faces, the difference in the facial features between the intact and Thatcherized faces is more recognisable in the upright faces than in the inverted faces.Given that the prediction error depends more on the perceptual than physical differences between the predicted and actual inputs under the predictive coding framework (Robinson, Breakspear, Young, & Johnston, 2020), we expected to observe a stronger oddball effect when the faces are presented upright than inverted.
We first presented intact and Thatcherized faces as standard and oddball stimuli, respectively, and compared the oddball effect when each was presented in the upright and inverted positions (Experiment 1).We then switched the faces between standard and oddball, i.e., Thatcherized and intact faces were presented as standard and oddball stimuli, respectively, and compared the oddball effect between the upright and inverted positions (Experiments 2 and 3).We also performed control experiments to examine whether there would be differences in perceived duration among intact and Thatcherized faces in the upright and inverted positions when they were presented without repeated presentation of the standard stimulus (Experiments 4 and 5).

Participants
A total of 12 paid volunteers (mean age ± SD = 21.2 ± 1.4) participated in Experiment 1.The sample size was pre-determined using G*power by computing the statistical power of significant difference between the conditions using paired t-test.The power analysis revealed that 12 was sufficient to detect the temporal oddball effect, namely greater duration expansion for upright Thatcherized face than inverted Thatcherized face with strong effect size (dz = 0.8), and a statistical power of 0.8 using a 5% significant level.The effect size was based on previous studies of the oddball effect (Sarodo et al., 2022;Saurels, Lipp, Yarrow, & Arnold, 2020).All participants had normal or corrected-tonormal vision, were naïve to the purpose of the experiment, and gave written informed consent.This study was approved by the internal review board of Waseda University.

Apparatus
The experiment was conducted in a darkened environment.Stimuli were presented on a gamma-corrected LCD monitor (1920 × pixels, 23.5-inch, and 100 Hz) controlled by a Mac mini.A chin rest restrained the participants' head movements at a viewing distance of cm from the display.The stimulus presentation code was written in Python with the PsychoPy toolbox (Peirce et al., 2019)

Stimuli
We used 15 images of Asian faces obtained from the Chicago face database (Ma, Correll, & Wittenbrink, 2015).All images were first cropped into ellipse shapes of equal size (500 × 400 pixels) and each face was then Thatcherized by inverting the mouth and the eyes using Adobe Photoshop.The intact faces were used as standard stimuli and the Thatcherized faces were used as the oddball stimulus.Fig. 1 shows an example of intact and Thatcherized faces in upright and inverted positions.

Procedure
A schematic illustration of the trial sequence is shown in Fig. 2. In each trial, participants observed five face stimuli presented at the centre of the screen and judged whether the duration of the last stimulus was longer or shorter than that of the penultimate stimulus.The faces subtended a visual angle of 5 • × 4 • .The first four stimuli were an identical intact face, randomly chosen from the 15 faces.The last stimulus was the Thatcherized version of the standard intact faces.In each trial, faces were presented either in an upright orientation or an inverted A. Sarodo et al. orientation.The orientation of the faces was randomly determined across trials.The interstimulus interval was 400 ms.The standard durations were 500 ms and the oddball duration was randomly chosen from the seven possible durations (350, 400, 450, 500, 550, 600, and 650 ms).After the oddball disappeared, participants indicated whether the duration of the last stimulus was longer or shorter than that of the penultimate stimulus by key press.The next trial automatically began 1000 ms after their key response.The participants were instructed not to count or tap to follow a rhythm throughout the experiment to avoid the temporal judgment based on cognitive strategies.
Prior to the experiment, participants were informed that a Thatcherized face would be presented at the fifth position of the sequence both in the upright and inverted conditions.There were two test blocks of 105 trials.Each participant completed 210 trials, giving 105 trials for each orientation condition and 15 responses for each oddball duration.Participants were required to take a 5-min rest between the blocks.

Analysis
We first calculated the point of subjective equality (PSE), which represents the oddball duration judged to be equal to the standard duration (500 ms).Sigmoid psychometric functions were fitted to the proportion of long responses for each participant and each orientation condition using the Psignifit toolbox for Python (Wichmann and Hill, 2001a,b).The 50 % point of the psychometric function was used as the PSE for the oddball stimulus.We then calculated the Relative Duration Distortion (RDD) by the equation RDD = (500 ms -PSE)/PSE × 100 to quantitatively evaluate the size of the oddball effect for each orientation condition (Cai et al., 2015).We also calculated the Weber ratio to analyse the temporal sensitivity to the oddballs.This was obtained by dividing the difference limen (DL), which is half of the difference between the 25 % and 75 % points of the psychometric function divided by the PSE.We performed paired t-tests to compare the RDD and Weber ratio between the orientation conditions.A significance threshold of p < 0.05 was chosen for all tests.In addition, we also performed a Bayesian statistical analysis on the data using JASP (Wagenmakers et al., 2018).We interpreted the results of Bayesian statistical analysis based on Jefferys (1961) interpretation of the Bayes factor.

Results & Discussion
Fig. 3a shows the psychometric function averaged across the participants.Fig. 3b represents the mean and individual RDD for each condition.One-sample t-tests with the Bonferroni correction revealed that RDD in the upright condition was significantly greater than zero (t (11) = 3.95, p = 0.002, dz = 1.14), but not in the inverted condition (t (11) = 1.15, p = 0.27, dz = 0.33).A paired-sample t-test showed that the RDD of the upright condition was significantly greater than that of the inverted condition (t(11) = 3.64, p = 0.002, dz = 1.05).A Bayesian  paired-sample t-test provided strong evidence (BF 10 = 25.71) for the alternative hypothesis that the RDD was greater in the upright condition than in the inverted condition.
The mean Weber ratios were 0.08 in the upright condition and 0.07 in the inverted condition.A paired-sample t-test revealed no significant difference between the upright and the inverted conditions (t(11) = 0.39, p = 0.71).A Bayesian paired-sample t-test provided moderate evidence for the null, compared to the alternative, hypothesis (BF 10 = 0.31).
The results of Experiment 1 revealed that the oddball effect was stronger in the upright condition, suggesting that the perceived duration of the upright oddball was longer than in the inverted condition.The Weber ratio was not different between the orientation conditions, indicating that the greater oddball effect was not due to the difference in the temporal sensitivity, nor was it a result of decisional biases such as central tendency (Hollingworth, 1910) or serial dependence (Wehrman, John, & Paul, 2020;Wehrman, Sanders, & Wearden, 2023).These results suggest that perceptual rather than physical difference between the repeatedly presented standard stimulus and the oddball stimulus contributes to the oddball effect.In the next experiment, we tested whether the results would be replicated when the temporal order of face types used as the standard and oddball stimuli is reversed.If the perceptual difference between the standard and oddball stimuli can alone account for the oddball effect, we would still observe difference in RDD between the orientation conditions even when the Thatcherized faces are repeated as the standard stimuli.

Participants
A total of 12 participants (mean age ± SD = 20.8 ± 1.5) were newly recruited and participated in Experiment 2. The sample size was predetermined to be the same as that of Experiment 1.

Stimuli, procedure, and analysis
The stimuli, procedure, and analysis were the same as those used in Experiment 1 except that the Thatcherized faces were presented as the standard stimuli and the intact faces were presented as the oddball stimulus.A schematic illustration of the trial sequence is shown in Fig. 4.

Results & Discussion
Fig. 5a shows the psychometric function averaged across the participants.Fig. 5b represents the mean and individual RDD for each condition.One-sample t-tests with the Bonferroni correction revealed that RDD was not significantly different from zero in any of the  orientation conditions (upright t(11) = 1.70, p = 0.12, dz = 0.49; inverted t(11) = 1.28, p = 0.23, dz = 0.37).A paired-sample t-test revealed that there was no significant difference in RDD between the upright and the inverted conditions (t(11) = 0.09, p = 0.46).A Bayesian paired-sample t-test also provided moderate evidence for the null hypothesis (BF 10 = 0.31) that there was no difference in the RDDs between the orientation conditions.
The mean Weber ratios were 0.10 for the upright oddball and 0.09 for the inverted oddball.A paired-sample t-test revealed no significant difference between the upright and the inverted conditions (t(11) = 0.91, p = 0.38).A Bayesian paired-sample t-test provided anecdotal evidence for the null hypothesis (BF 10 = 0.41).
In contrast to Experiment 1, the results of Experiment 2 revealed no difference in the oddball effect between the upright and inverted conditions, even though the method was identical except that the face stimuli used as the standard and oddball were interchanged.This suggests that reversing the standard and oddball stimuli can produce a different influence on the oddball effect.However, the disappearance of the stronger oddball effect may be due to the larger individual differences because the variance of the RDDs among participants was greater for Experiment 2 than for Experiment 1. Indeed, Bartlett's test for homoscedasticity revealed unequal variance for the inverted conditions between Experiments 1 and 2 (p = 0.003).To address this possibility, we conducted Experiment 3 where the same group of participants performed the conditions in both Experiment 1 and Experiment 2.

Participants
A total of 18 participants (mean age ± SD = 21.2 ± 3.7) were newly recruited and participated in Experiment 3. Data from one participant was excluded due to poor performance (Weber ratio above 0.2).The final sample size of 17 was pre-determined using PANGEA (Westfall, 2016) by computing the statistical power of the interaction effect between the sequence and orientation in ANOVA.Because there was no report on the estimation of the effect size of the interaction effect between the sequence and orientation, we decided to set the effect size medium for the power analysis.The power analysis revealed that 17 were sufficient to detect the interaction with the medium effect size (d = 0.45) and a statistical power of 0.8.We recruited participants until the number of valid data reached the pre-determined sample size.

Stimuli, procedure, and analysis
The stimuli, procedure, and analysis were the same as those used in Experiment 1 and Experiment 2 except for the following.In the intactrepeated condition, intact faces were presented as the standard stimuli, and the Thatcherized faces were presented as the oddball stimulus, whereas, in the Thatcherized-repeated condition, the Thatcherized faces were presented as the standard stimuli and the intact faces were presented as the oddball stimulus.The stimulus sequence was randomly determined across trials.We performed a 2 × 2 repeated measures ANOVA with factors sequence and orientation.
We additionally analysed the combined data of 41 participants obtained in Experiments 1, 2, and 3 1 .Fig. S1 shows the mean and individual RDD for each condition.One-sample t-tests with the Bonferroni correction revealed that the RDD was significantly greater than zero in all the conditions (ps < 0.05).We further performed a linear mixed effect model analysis on the RDD with the orientation and sequence as fixed effects variables and the participants as a random effects variable.The results revealed that orientation (F(1, 72.51) = 9.79, p = 0.003), but not the sequence (F(1, 81.44) = 1.46, p = 0.23) predicted the difference in the RDD.Importantly, the interaction between the orientation and the sequence (F(1, 72.51) = 4.05, p = 0.048) also predicted the difference in the RDD.
Unexpectedly, the combined analysis revealed that the oddball effect occurred even in the inverted conditions, where there was little physical and perceptual difference between the standards and oddball.Wehrman, Wearden, and Sowman (2020b) demonstrated that an oddball induces stronger duration distortion when the temporal position of the oddball in the sequence is predictable, suggesting that the positional expectation increases perceived duration.Given that the oddball was 1 These combined analyses were not originally planned and conduced as post hoc analysis following the reviewer's suggestion.
A. Sarodo et al. always presented at the fifth position in our study, such temporal expectation may have induced the oddball effect in the inverted conditions.This is consistent with previous studies showing that even repeated stimuli can induce duration dilation when the temporal position of the last stimulus in the sequence is predictable (Birngruber, Schröter, & Ulrich, 2015b;Saurels, Yarrow, Lipp, & Arnold, 2023).
The comparison of the RDDs among conditions revealed that the orientation of the stimuli influenced the oddball effect only when the Thatcherized face was presented as the oddball stimulus after the repetition of the intact standard faces.These results are consistent with those of Experiments 1 and 2, suggesting that the difference between the intact-repeated and Thatcherized-repeated conditions was replicable even when the same participants completed both conditions.Furthermore, the linear mixed model revealed that the interaction between the orientation and sequence predicted the difference in the RDDs providing stronger evidence for these asymmetric influences of perceptual difference.However, it may be also possible that the stronger oddball effect for the intact-repeated condition is due to the difference in perceived duration between the intact and Thatcherized faces themselves in the upright orientation.Because emotional salience can expand the perceived duration of face stimuli (Droit-Volet et al., 2004), the Thatcherized face itself might have been perceived as longer in duration than the intact face due to its oddness, which could result in the stronger oddball effect when the Thatcherized face was used as the oddball.To test this possibility, in Experiment 4, we compared perceived durations of the intact and Thatcherized face against a reference stimulus both in the upright and inverted orientations.

Participants
A total of 19 participants (mean age ± SD = 20.4 ± 1.8) were newly recruited and participated in Experiment 4. Data from two participants was excluded due to poor performance (Weber ratio above 0.2).The final sample size of 17 was pre-determined as in Experiment 3. We recruited participants until the number of valid data reached the predetermined sample size.

Stimuli and Procedure
The experiment was divided into four blocks of 105 trials.In a trial, a reference stimulus and a comparison stimulus were subsequently presented with an ISI.In each trial, the participants observed a pair of stimuli.The reference stimulus was a beige oval subtending a visual angle of 5 • × 4 • .The comparison stimulus was an upright-intact, upright-Thatcherized, inverted-intact, and inverted-Thatcherized face (Fig. 7).The face type of the comparison stimulus was fixed within a block.The duration of the reference stimulus was 500 ms, the comparison duration was randomly chosen from one of the seven possible durations (350, 400, 450, 500, 550, 600, and 650 ms), and the ISI was 400 ms.Participants were asked to compare the durations of the reference and comparison stimuli and indicate whether the duration of the comparison stimulus was longer or shorter than that of the reference stimulus.The order of the blocks was randomised across participants.

Analysis
We calculated the RDD to quantitatively evaluate the size of the PSE shift from the standard duration for each condition.We performed a 2 × 2 repeated measures ANOVA with factors face type and orientation.
The results of Experiment 4 demonstrated that the perceived duration of the faces did not differ regardless of face type and orientation.This suggests that the emotional salience elicited by the Thatcherized faces is not sufficient to influence the perceived duration and excludes the possibility that the different oddball effects observed in previous experiments are primarily due to the difference in perceived duration of the face stimuli used as the oddball.However, it is still possible that the emotional salience of the Thatcherized faces was enhanced in Experiment 1 because they were preceded by the intact faces.To address this possibility, in Experiment 5, we measured the perceived duration of the intact and Thatcherized faces by using the intact face as the reference stimulus.

Participants
A total of 23 participants (mean age ± SD = 20.3 ± 2.2) were newly recruited and participated in Experiment 5. Data from six participants was excluded due to poor performance (Weber ratio above 0.2).The sample size of 17 was pre-determined to be the same as in Experiment 4. We recruited participants until the number of valid data reached the predetermined sample size.

Stimuli, procedure, and analysis
The stimuli, procedure, and analysis were the same as those used in Experiment 4 except that the intact faces were used as the reference stimuli (Fig. 9).In the upright conditions, the reference stimuli were upright intact faces, whereas, in the inverted conditions, the reference stimuli were inverted intact faces.
The results of Experiment 5 demonstrated that the perceived duration of the faces did not differ regardless of face type and orientation conditions even when the intact face was used as the reference stimulus.It is worth noting that the stimuli and procedure in the Thatcherized conditions were the same as in Experiment 1 except that the intact face was presented once, suggesting that the repeated presentation of the intact face was important for the difference in RDD observed in Experiment 1. Pariyadath & Eagleman (2012) showed that the oddball effect did not occur when there was only one standard presentation before the oddball stimulus.Thus, the result of Experiment 5, together with those of Experiment 1, imply that the difference in the RDDs actually reflects the difference in the oddball effect.
Although we repeated an identical face in the intact face conditions, we did not observe a reduction in the perceived duration of the comparison face, which is inconsistent with the previous studies showing that repetition of identical stimulus reduces the perceived duration (Birngruber et al., 2015a;Matthews, 2011;Noguchi & Kakigi, 2006).Because we fixed the temporal position of the comparison stimulus throughout the experiment, it is possible that positional expectation (Wehrman et al., 2020b) cancelled out the repetition reduction effect.

General Discussion
In the present study, we used the Thatcher illusion to examine whether the perceptual or physical difference between the repeated and the oddball stimuli is important for the oddball effect.Because the difference in local features of a face is less likely to be detected when the whole face is inverted, we predicted that the oddball effect would be stronger when the stimuli were presented upright than inverted, if the perceptual difference is important for the oddball effect.We first presented the intact and Thatcherized faces as the standard and oddball stimuli, respectively, and found a stronger oddball effect when the faces were presented in the upright position than when they were presented in the inverted position (Experiment 1).However, the difference in the oddball effect was not observed when the standard and oddball stimuli were interchanged, i.e., the intact face was presented as the oddball after repeated presentation of Thatcherized faces (Experiments 2).These results were replicated when the same participants completed both the intact-repeated and Thatcherized-repeated conditions (Experiment 3).Control experiments using a two-interval discrimination task confirmed that the perceived duration of the face stimuli was not different across face type and orientation when they were presented without repeated presentation of the preceding stimuli (Experiments 4 and 5).Taken together, the results of the present study suggest that the perceptual difference between the repeated and the oddball stimuli may play a role, but cannot fully account for the oddball effect by itself.
Although the stimuli used were identical between the intact-repeated and Thatcherized-repeated conditions, the influence of face orientation on the oddball effect was found only in the intact-repeated condition.This indicates that whether the perceptual difference affects the oddball effect depends on how the stimuli are presented, i.e., which of the intact or Thatcherized faces is presented as the standard (or oddball) stimulus.Neurophysiological studies demonstrated that a repetition of unfamiliar stimuli induces weaker repetition suppression than that of familiar stimuli (Fiebach, Gruber, & Supp, 2005;Henson, Shallice, & Dolan, 2000;Vizioli, Rousselet, & Caldara, 2010).In the Thatcherized-repeated condition, the locally inverted unfamiliar faces were presented  repeatedly and thus may have produced smaller repetition suppression.This is consistent with our previous finding that after the repeated presentation of a human face, the oddball effect became stronger when the oddball was a cat face than when it was a novel human face, whereas the influence of the change in face category disappeared when the cat face (i.e., unfamilir face) was repeatedly presented as the standard (Sarodo et al., 2022).
The present results also have some implications for the predictive coding account of the oddball effect (Pariyadath & Eagleman, 2012;Schindel et al., 2011).Repetition suppression is suggested to derive from increased predictability regarding the repeated stimuli (see Grotheer & Kovács, 2016, for a review).In the predictive coding framework, the prediction regarding the forthcoming stimulus is represented as the probability distribution based on the weighted average of the prior sensory information (Knill & Pouget, 2004).Repetition of unreliable stimuli, such as stimuli with no prior information or noise, forms predictions with low reliability and high variance, resulting in a smaller prediction error even when an oddball appears.If this is the case, the present results suggest that the perceptual difference contributes to the oddball effect via prediction error and that repetition of a reliable stimulus, such as a familiar face, may be important to induce sufficient prediction error from the perceptual difference.
While the predictive coding is a plausible account of the oddball effect, a growing body of evidence suggests that the influence of predictability on perceived duration is not constant, and it can vary based on how the prediction is established (Birngruber, Schröter, Schütt, & Ulrich, 2018;Cai, Eagleman, & Ma, 2015;Matthews, 2015;Saurels et al., 2022;Schweitzer, Trapp, & Bar, 2017;Skylark & Gheorghiu, 2017;Warda & Khan, 2022;Wehrman et al., 2020b).For example, Matthews (2015) presented a pair of stimuli in which the second stimulus is a repeat of the first stimulus or a novel stimulus and found that the repetition contraction disappears when the repeat stimulus was more frequently presented than the novel stimulus.At the neural level, neural suppression induced by repetition probability (expectation suppression) is dissociated from repetition suppression (Grotheer & Kovács, 2015;Todorovic & de Lange, 2012).Repetition suppression is formed by the prediction based on prior stimulus repetition and believed to be rooted in the sensory cortex, whereas the expectation suppression is suggested to originate from the higher cognitive processing at the frontal cortex (Grotheer & Kovács, 2015;Summerfield et al., 2006).Thus, we believe that the difference in the perceived duration between the orientation conditions is primarily caused by the violation of repetition prediction at the perceptual level.This is consistent with the previous finding that the influence of repetition probability is eliminated when repetition prediction is violated within the sequence, suggesting that the violation of repetition prediction is a primary cause of the oddball effect (Warda & Khan, 2022).
Attention is also suggested to play a role in the oddball effect.Several researchers have argued that the oddball effect is caused by an increased attention allocation to oddball stimuli compared to repeated stimuli (New & Scholl, 2009;Tse et al., 2004;van Wassenhove, Buonomano, Shimojo, and Shams, 2008).Thus, the difference in the attentional allocation to the intact and Thatcherized faces may account for the discrepancy in the oddball effect between the intact-repeated and Thatcherized-repeated conditions.However, Pariyadath and Eagleman (2007) demonstrated that the size of the oddball effect did not differ between neutral and emotionally salient oddballs, suggesting that the amount of attention to the oddball plays a minimal role in the oddball effect.Although these previous studies focused on the salience of the oddball, future studies should also address how the salience of repeated stimuli affects the oddball effect, for example, by using neutral oddball and salient repeated stimuli.
In conclusion, the results of the present study revealed that the oddball effect depends more on the perceptual than physical difference between the standard and oddball stimuli.This supports the previous findings showing that perceptual rather than physical aspects of visual stimuli contribute more to the duration distortion (Aoki, Kawano, Terao, & Murakami, 2016;Ono & Kawahara, 2007;Yamamoto & Miura, 2016).Furthermore, we found that the influence of the perceptual difference was not observed when unfamiliar stimuli were presented repeatedly before presenting the oddball.We believe that this asymmetric influence reflects the different prediction errors for the oddball stimulus and that the predictive coding process plays an important role in the oddball effect.

Fig. 1 .
Fig. 1.Intact and Thatcherized faces in the upright and inverted positions.

Fig. 2 .
Fig. 2. A schematic diagram depicting the basic trial sequence of the temporal oddball task in Experiment 1.The participants compared the durations of the last and the penultimate stimuli.

Fig. 3 .
Fig. 3. Results of Experiment 1.(a) The participants-averaged psychometric functions for each condition.(b) The mean RDD of the last stimulus against the duration of the penultimate stimulus and individual RDDs in each condition (* p < 0.05, ** p < 0.01, *** p < 0.001).Error bars indicate the standard error of the mean.

Fig. 4 .
Fig. 4. A schematic diagram depicting the basic trial sequence of the temporal oddball task in Experiment 2. The participants compared the durations of the last and the penultimate stimuli.

Fig. 5 .
Fig. 5. Results of Experiment 2. (a) The participants-averaged psychometric functions for each condition.(b) The mean RDD of the last stimulus compared to that of the penultimate stimulus and individual RDDs in each condition.Error bars indicate the standard error of the mean.

Fig. 6 .
Fig. 6.The mean RDD of the last stimulus compared to that of the penultimate stimulus and individual RDDs in each condition.Error bars indicate the standard error of the mean.

Fig. 7 .
Fig. 7.A schematic diagram depicting the basic trial sequence of Experiment 4. The participants compared the durations of the first (i.e., reference) and second (i.e., comparison) stimuli.

Fig. 8 .
Fig. 8.The mean RDD of the comparison stimulus compared to that of the standard stimulus and individual RDDs in each condition.Error bars indicate the standard error of the mean

Fig. 9 .
Fig. 9.A schematic diagram depicting the basic trial sequence of Experiment 5.The participants compared the durations of the first (i.e., reference) and second (i.e., comparison) stimuli.

Fig. 10 .
Fig. 10.The mean RDD of the comparison stimulus compared to that of the reference stimulus and individual RDDs in each condition.Error bars indicate the standard error of the mean.