SEARCH

SEARCH BY CITATION

Keywords:

  • event-related potentials;
  • perception;
  • visual attention

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We investigated the sensitivity of visual mismatch negativity (vMMN) to an abstract and non-semantic category, vertical mirror symmetry. Event-related potentials (ERPs) elicited by random and symmetric square patterns, delivered in passive oddball paradigm (participants played a video game), were recorded. In one of the conditions, symmetric patterns were frequent (standard) stimuli and the random patterns were infrequent (deviant) stimuli; in the other condition, the probabilities were reversed. We compared the ERPs elicited by symmetric stimuli as deviants and as standards, and, similarly, the ERPs elicited by the random deviants and random standards. As the difference between the ERPs elicited by random deviant and random standard stimuli, a posterior negativity emerged in two latency ranges (112–120 and 284–292 ms). These negativities were considered to be vMMN components. We suggest that the two vMMN components are organised in cascade error signals. However, there was no significant difference between the ERPs elicited by symmetric deviants and those elicited by symmetric standards. The emergence of vMMN in response to the deviant random stimuli is considered to be a deviation of a perceptual category (in the symmetric standard sequence presented). Accordingly, random stimuli acquired no perceptual category; for this reason, the symmetric deviant (in the random standard sequence presented) elicited no vMMN. The results show that the memory system underlying vMMN is capable of coding perceptual categories such as bilateral symmetry, even if the stimulus patterns are unrelated to the ongoing behavior.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

At the level of conscious experience, the visual system is surprisingly insensitive to environmental changes if such changes are outside the focus of attention (Simons & Levin, 1997). However, research on the visual mismatch negativity (vMMN) component of event-related potentials (ERPs) shows that non-attended visual changes violating the regularity of stimulation are registered in posterior brain structures. In fact, vMMN occurs even if participants cannot report the stimulus change (Czigler & Pató, 2009) or the change appears during a period of attentional blink (Berti, 2011).

Visual mismatch negativity (an ERP component in the 100–300-ms latency range) is a counterpart of auditory mismatch negativity [for reviews, see Kujala et al. (2007) and Näätänen et al. (2007)]. vMMN is elicited by various deviant visual features, such as color (Czigler et al., 2002), orientation (Astikainen et al., 2008), movement direction (Pazo-Alvarez et al., 2004), spatial frequency (Heslenfeld, 2003), and contrast (Stagg et al., 2004). Besides being sensitivite to single visual features, the system underlying vMMN is sensitive to more complex visual changes, such as deviant conjunction of visual features (Winkler et al., 2005) and deviant sequential relationships (Stefanics et al., 2011); for reviews, see Czigler (2007) and Kimura et al. (2011). Some ERP studies have shown that vMMN is sensitive to stimulus categorisation in the case of facial expressions (Astikainen & Hietanen, 2009; Stefanics et al., 2012). Categorical sensitivity in the color domain has also been demonstrated.

Clifford et al. (2010) and Mo et al. (2011) reported that larger vMMNs were elicited if deviants and standards belonged to different color categories than if deviants and standards belonged to the same color category, irrespective of whether the distances in color space were identical.

The present study aimed to investigate whether the implicit system underlying vMMN was capable of registering vertical mirror symmetry as a perceptual category. Several behavioral studies have shown that the visual system is particularly sensitive to various forms of symmetry (for a review, see Treder (2010)). According to Carmody et al. (1977) and Tyler et al. (1995), stimulus duration in the 40–80-ms range is long enough for the recognition of symmetric patterns. Other behavioral studies have shown that symmetry can be detected automatically (Baylis & Driver, 1994; Wagemans, 1995; Huang et al., 2004; Machilsen et al., 2009).

Vertical mirror symmetry is a salient feature of living objects, and has obvious biological significance (Tyler & Hardage, 1996). However, so far, no ERP study has analysed the level of processing that is sensitive to symmetry and the automaticity of sensitivity to symmetry.

Few studies have investigated the processing of symmetric stimuli on the basis of event-related brain activity. Jacobsen & Höfel (2003) and Höfel & Jacobsen (2007) reported a posterior negative wave elicited by symmetric patterns. In these studies, symmetry as such was task-irrelevant; participants made aesthetic judgements, performed a detection task, or contemplated the beauty of the stimuli. The negativity emerged in the 380–890-ms poststimulus latency range, so this effect may not be a correlate of elementary perceptual processes. However, in a sequence of alternatively presented random and symmetric dot-patterns, the symmetric patterns elicited a sustained posterior negativity with ~ 220-ms onset, whereas random patterns elicited positivity with earlier onset (~ 130 ms) (Norcia et al., 2002). Such activities were considered to be correlates of the appearance of global forms, i.e. an activity more general than a specific symmetry effect.

In the present study, we tested whether the system underlying vMMN is sensitive to symmetry as a perceptual category. If this is so, the regular presentation of stimuli belonging to the same perceptual category (symmetry) will establish a mental representation containing the sequential rule of the stimulation. Irregular stimuli (which do not belong to this category) will violate the prediction that derives from mental representation, and therefore elicit the vMMN component. For this reason, we infrequently embedded symmetric patterned stimuli (deviants) in a series of random patterned stimuli (standards), whereas, in another condition, random deviants appeared in the context of symmetric standards. Thus, we could compare the ERPs elicited by categorically identical standard and deviant stimuli. We expect an ERP difference between the deviant and standard random pattern; we hypothesise that the ERP difference is a vMMN, i.e. a posterior negativity within the 100–300-ms latency range. However, ‘randomness’ cannot be considered as a categorical rule, so a deviant symmetric pattern does not violate a perceptual regularity. We suggest that no similar difference is expected in the case of symmetric deviants.

The vMMN-related stimuli – high-contrast black-and-gray squares – were presented on the lower half of the visual field, as the lower half-field stimulation usually elicits more pronounced ERP components (Jeffreys & Axford, 1972) and vMMN (Sulykos & Czigler, 2011). The task-related stimuli were delivered on the opposite half of the visual field. The visual task required continuous fixation on the center of the task-field.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Participants

Participants were 12 paid students (four women; mean age, 21.8 years; standard deviation, 1.7 years) with normal or corrected-to-normal vision. Written consent was obtained from all participants prior to the experimental procedure. The study was conducted in accordance with the Declaration of Helsinki, and approved by the United Committee of Ethics of the Psychology Institute in Hungary.

Stimuli

The stimuli were either bilaterally symmetric or random black-and-gray square patterns. Patterns with vertical symmetry were used, because this type of symmetry is more prominent than horizontal symmetry (Barlow & Reeves, 1979; Wagemans et al., 1991). The size of a square item was 1° from the 1.2-m viewing distance. The pattern consisted of two matrices of 16 items (four columns and four rows); therefore, the size of the pattern was 4° × 4° in each half-field. The two halves of the pattern were separated by a vertical line of 0.3°, and the task-field and the patterns were separated by a horizontal line of 0.4°. Each matrix consisted of nine gray squares and seven black squares. Figure 1 shows a sample stimulus (A) and the experimental stimulus sequences (B).

image

Figure 1. Stimuli and illustration of the sequence applied. (A) A schematic illustration of the presented pattern and the task. (B) The two different probability conditions used in the experiment.

Download figure to PowerPoint

The luminance of the gray squares was 20.1 cd/m2, and the (Weber) contrast was 3.54. The stimuli appeared on a 17-inch monitor (Samsung SyncMaster 740B; 60-Hz refresh rate) in a dimly lit and soundproof room. The stimulus duration was 167 ms, and the interstimulus interval was 417 ms. Before the repetition of a particular pattern, at least four physically different patterns were presented. Symmetric and random stimuli were delivered in oddball sequences. In one of the conditions, symmetric patterns were the frequent (standard) stimuli (P = 0.84) and random patterns were the deviant stimuli (P = 0.16). In the other condition, these probabilities were reversed; that is, the random patterns were standards, and the rare symmetric patterns were deviants. A sequence consisted of 400 stimuli. There were two sequences for both conditions. The sequences were delivered in alternate order (ABAB or BABA). The sequence orders were counterbalanced across participants.

Task

The stimuli for the task appeared on the upper half of the visual field (Fig. 1). To facilitate the participants' interest, the primary task was designed as a simple video game. The participants used a gamepad controller to maneuver a spaceship in a canyon. The canyon – a rectangular cross-section tube – lay in the surface of a schematic planet. In the canyon, there were three types of spaceship marked by different colors (blue, red, and green). The color of the controlled spaceship was blue. That was directed with the gamepad along the horizontal dimension of the canyon. In every second, one spaceship appeared at the start of the canyon and moved towards the blue spaceship. The color of the spaceship was red with 0.6 probability and green with 0.4 probability. The aim of the task was to avoid the red spaceships and to catch the green ones with the controlled spaceship. To perform the task properly, participants had to fixate in the location where the spaceships appeared. For more details, see Sulykos & Czigler (2011).

Measuring brain electrical activity

Electroencephalographic activity was recorded (DC, 70 Hz; sampling rate, 500 Hz; Synamps2 amplifier, NeuroScan recording system) with Ag/AgCl electrodes placed at 61 locations according to the extended 10–20 system by use of an elastic electrode cap (EasyCap). The reference electrode was on the nose tip, and offline re-referenced to the average activity.‎ Horizontal electrooculographic activity was recorded with a bipolar configuration between electrodes positioned lateral to the outer canthi of the eyes. Vertical eye movement was monitored with a bipolar montage between electrodes placed above and below the right eye.

The electroencephalographic signal was bandpass-filtered offline, with cutoff frequencies of 0.1 and 30 Hz (24-dB slope). Epochs of duration 600 ms, including a 100-ms prestimulus interval, were extracted for each event, and averaged separately for the standard and deviant stimuli. The mean voltage during the 100-ms prestimulus interval was used as the baseline for amplitude measurements, and epochs with an amplitude change exceeding ± 50 μV on any channel were excluded from further analysis.

Event-related potentials were averaged separately for the standard and deviant stimuli (symmetric and random) in the two conditions. Responses to the third to the seventh standards after a deviant were included in the standard-related ERPs. To identify change-related activities, ERPs elicited by standard stimuli were subtracted from ERPs elicited by deviant stimuli in the reverse condition.

Note that, in many studies, vMMN was calculated as the difference between the ERPs elicited by deviant and standard of the same stimulus sequence. With this method, the effect of physical differences between the deviant and standard and the effect of memory-related mismatch effects are confounded. Therefore, comparison of ERPs elicited by identical stimuli is highly recommended (Kujala et al., 2007). Furthermore, comparison of physically identical stimuli (presented frequently/infrequently) in different conditions will not be sufficient to get rid of refractoriness effects adding to plain memory-related effects (Kimura et al., 2009). However, this problem does not apply to our study, as we used different types of standard stimuli, making a contribution of refractoriness effects to our vMMN response rather unlikely.

Visual mismatch negativity was identified if, within the 100–300-ms latency range, deviant-minus-standard amplitude difference was different from zero at least at five subsequent points at any occipital location [for reviews of the characteristics of the range and surface distribution of the vMMN, see Czigler (2007) and Kimura (2011)]. In this way, we identified an earlier (112–120 ms) and a later (284–292 ms) range of the difference potentials. At six electrode locations (PO3, POz, PO4, O1, Oz, and O2) as regions of interest, the average amplitude values of these epochs were calculated, and entered into anovas with factors of probability (deviant or standard), anteriority (parieto-occipital or occipital), and laterality (left, midline, or right). We compared, at the same electrode locations, the peak latencies and scalp distributions of the exogenous components and the difference potentials.

Note that, at lower half-field stimulation, the C1 and C3 components are positive and the C2 component is negative. Investigation of the relationship between a negative component and the vMMN is relevant, because it is important to separate the refractoriness/habituation of an exogenous activity from vMMN. In this context, the similar analysis of the positive components (C1 and C3) is less important, because reduced exogenous positivities elicited by the deviant stimuli cannot be expected (in the case of stimulus-specific refractoriness/habituation, amplitude reduction is expected, i.e. positive deviant-minus-standard difference).

Peak latencies were measured at the maxima of the components. The distributions of the difference potential and the C2 component were compared with vector-scaled amplitude values (McCarthy & Wood, 1985). Where appropriate, Greenhouse–Geisser correction was applied. Effect size was characterised as partial eta-squared (η2). Post hoc analyses were performed with Tukey's HSD test. In the reported effects, the alpha level was at least 0.05.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Behavioral results

Participants avoided the red ship with a frequency of 82% (standard error of the mean, 1.53%), and caught the green ship with a frequency of 83% (standard error of the mean, 1.05%). This difference was not significant. There was no also difference in performance between the random and symmetric standard conditions.

Event-related potentials

Figure 2 shows the ERPs elicited by the symmetric (A) and random (B) stimuli, as both standards and deviants, and also the deviant-minus-standard difference potentials. The stimuli elicited a positive–negative–positive (C1–C2–C3) set of pattern-specific exogenous components (Jeffreys & Axford, 1972). Table 1 shows the latency values of the exogenous components, and Fig. 3 shows the scalp distribution of the C1, C2 and C3 components and the difference surface distributions. Figure 2 shows that the deviant and standard symmetric stimuli elicited similar ERPs. In fact, in the t-tests, the difference did not reach the criterion level (deviant-minus-standard amplitude difference is different from zero at at least five subsequent points). However, over the posterior–occipital locations, random deviant and random standards were different in an earlier (112–120 ms) and in a later (284–292 ms) range. In both ranges, the difference was negative. Table 2 shows the amplitudes of the random deviants and standards in the two ranges.

image

Figure 2. Comparison of the categorically identical stimuli. Group average ERPs elicited by the standard and deviant stimuli, and difference potentials. (A) Responses to symmetric stimuli. (B) ERP responses to random stimuli. The random deviants elicited negativities in two latency ranges (earlier, 112–120 ms; later, 284–292 ms).

Download figure to PowerPoint

image

Figure 3. Topographic maps of the random and symmetric exogenous components (C1, C2 and C3 scalp distributions at the peak latencies) and the surface distributions of the deviant-minus-standard difference potentials in the three ranges.

Download figure to PowerPoint

Table 1. Latency values (ms) of the pattern-specific event-related components (Standard error of the mean in parenthesis)
 C1C2C3
SymmetricRandomSymmetricRandomSymmetricRandom
Oz86.16 (2.69)86.83 (2.97)124.8 (4.64)125.2 (4.70)243.3 (3.43)242.0 (3.67)
POz88.00 (2.17)86.66 (1.98)130.0 (3.32)130.5 (3.52)240.3 (3.02)239.8 (2.87)
Table 2. Mean amplitude values (μV) of the event-related potentials elicited by random deviants and standards (Standard error of the mean in parenthesis)
 112–120ms284–292 ms
StandardDeviantStandardDeviant
Oz−1.13 (0.47)−1.71 (0.31)1.67 (0.29)1.35 (0.30)
POz−0.96 (0.48)−1.46 (0.45)2.40 (0.48)2.28 (0.49)

Event-related potential amplitudes elicited by the deviant and standard random stimuli were compared in both latency ranges by the use of anovas with factors probability (deviant and standard), anteriority (parieto-occipital and occipital) and laterality (left, midline, and right). In the 112–120-ms range, only the probability main effect was significant (F1,11 = 6.31, P < 0.05, η2 = 0.36), showing the occipital/parieto-occipital distribution of the early negativity. In a similar analysis of the 284–292-ms range, the main effect of anteriority (F1,11 = 7.13, P < 0.05, η2 = 0.39) and the probability × anteriority interaction (F1,11 = 7.52, P < 0.05, η2 = 0.41) were significant. According to the Tukey HSD tests, the deviant-minus-standard difference was significant only at the occipital locations (P < 0.01 in all cases). As the results show, vMMN appeared in two latency ranges. However, it is possible that, instead of the emergence of vMMN, the earlier effect was an amplitude modulation of the C2 component. Nevertheless, as Fig. 2 shows, the latency of the difference potential was shorter at the occipital locations. To investigate the latency difference (116 vs. 130 ms), we compared the C2 and difference potential latencies at the parieto-occipital and occipital locations (POz and Oz). In an anova, the main effect of anteriority was significant (F1,11 = 6.33, P < 0.05, η2 = 0.36) and the component (difference vs. standard) × anteriority interaction was significant (F1,11 = 4.93, P < 0.05, η2 = 0.30). However, the main effect of component was only marginally significant (F1,11 = 3.46, P < 0.09, η2 = 0.24). To further investigate the relationship between the C2 and the difference potential, we compared the surface distributions. As Fig. 3 indicates, the distribution of the difference potential was wider than the C2 distribution. To investigate the possibility of distribution difference, we added further electrodes to both sides on both rows (P7, P8, PO7, and PO8) to the previous electrode set (PO3, POz, PO4, O1, Oz, and O2), and vector-scaled the data (McCarthy & Wood, 1985). The C2 amplitude was measured as the average of a ± 4-ms point around the peak of the component (130 ms). In an anova with factors component (C2 and difference potential), anteriority and laterality, only the three-way interaction was significant (F4,44 = 3.82, P < 0.05, ε = 0.53, η2 = 0.26). According to the Tukey HSD test, C2 was larger at the anterior row, and C2 amplitude was larger at the midline. We found significant differences in the distribution of early vMMN and C2. Additionally, we compared the vector-scaled amplitude values of the two vMMNs in an anova with factors difference potential (early and late) anteriority (parieto-occipital and occipital), and laterality (left, midline, and right). Owing to the lack of significant effects, we could not conclude that the surface distributions were different.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Frequent (standard) and infrequent (deviant) symmetric patterns elicited identical ERPs. However, in the context of symmetric patterns, random deviant stimuli elicited two posterior negative components. The negative difference potentials cannot be explained as the refractoriness of low-level visual processes, for the following reasons. First, the scalp distribution of the exogenous activity (C2 component) differed from the characteristics of the difference potential in the earlier latency range. Second, there was a tendency for there to be peak latency differences between the C2 and the difference potentials. Third, in the later latency range, there was no exogenous difference corresponding to the posterior negativity. We consider the two difference potentials as sub-components of vMMN. The emergence of multiple vMMNs is not unprecedented (Maekawa et al., 2005; Astikainen & Hietanen, 2009; Sulykos & Czigler, 2011). Considering the difference potentials as vMMN, we interpreted the asymmetry of the random and symmetry conditions as a manifestation of a category effect. Unlike the random patterns, symmetric stimuli may acquire a category. Rare random (deviant) stimuli violated the representation of the category (symmetry) and elicited vMMN. Thus far, category influences on vMMN have been reported in the color domain (Athanasopoulos et al., 2010; Clifford et al., 2010; Mo et al., 2011) and in the case of facial emotions (Zhao & Li, 2006; Astikainen & Hietanen, 2009; Stefanics et al., 2012). According to the present results, high-order visual features acquired a category without the involvement of attentional processes, and stimuli deviating from the sequential appearance of patterns belonging to such a category were automatically registered.

The present findings are in line with behavioral results showing the fast and automatic sensitivity of the visual system to symmetry (Carmody et al., 1977; Baylis & Driver, 1994; Tyler et al., 1995; Wagemans, 1995; Huang et al., 2004). According to some studies, short-latency vMMN is generated in retinotopic areas (Czigler et al., 2004; Pazo-Alvarez et al., 2004; Sulykos & Czigler, 2011). Nevertheless, according to neuroimaging and transcortical magnetic stimulation data, the loci of sensitivity to symmetry are above the retinotopic (i.e. V1 and V2) structures (Sasaki et al., 2005; Tyler et al., 2005; Cattaneo et al., 2011). An early effect of symmetry on ERPs was reported by Norcia et al. (2002); however, neither the patterns nor the stimulus presentation methods in that study were comparable to the methods used in the present study. Considering both the early and the late negativities as vMMNs, emergence of the successive components suggests a cascade of memory-related processes. This possibility fits the idea that mismatch responses are correlates of hierarchically organised error signals; i.e. the difference between a model predicting the characteristics of ongoing stimulation and bottom-up processes elicited by the actual stimulation (Winkler & Czigler, 2012). vMMNs in the earlier and later latency ranges had similar surface distributions. Therefore, it is unlikely that the early and late vMMNs are attributable to the structural hierarchy of the visual system. Instead, we consider the later component to be a manifestation of recurrent activity. So far, there have been only a few attempts to localise vMMNs. These studies identified the prestriate cortex as generator of vMMN (Czigler et al., 2004; Kimura et al., 2010; Sulykos & Czigler, 2011). According to a magnetoencephalography study, the middle occipital gyrus is an important cortical area whose activity reflects the sensory memory-based visual change-detection processes (Urakawa et al., 2010). Furthermore, Yucel et al. (2007) reported a deviant-related extensive network (occipital–fusiform, posterior parietal, prefrontal and subcortical regions). In these regions, unattended deviants elicited blood oxygen level-dependent activation that decreased with the difficulty of a demanding visuomotor tracking task.

The emergence of vMMN elicited by random deviants and the lack of vMMN elicited by symmetric deviants are analogous to an effect in auditory modality. Within a series of legal syllables in a language, an irregular syllable elicited mismatch negativity, but a legal deviant in a series of irregular ones did not (Steinberg et al., 2011). Accordingly, violation of an existing category resulted in automatic detection processes; however, in the absence of categorisation, there were no such processes. It seems that the role of category-related representation in the two modalities is similar.

In conclusion, the results of the present study show that bilateral vertical symmetry is a prominent stimulus category, and that stimuli violating the rule of successive appearance of such patterns elicit deviant-related components, even if the stimulus patterns are unrelated to the ongoing behavior.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

This work was supported by the Hungarian Research Found (OTKA 71600). We thank Professor John Foxe for helpful comments and suggestions.

Abbreviations
ERP

event-related potential

vMMN

visual mismatch negativity

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
ejn12061-sup-0001-VideoS1.avivideo/avi1082KVideo. S1. The illustration of the experimental display. The vMMN-related stimuli were delivered on the lower half of the visual field. The task was, to maneuver the blue spaceship among another spaceships, on the upper half of the visual field.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.