Elsevier

Consciousness and Cognition

Volume 21, Issue 3, September 2012, Pages 1222-1231
Consciousness and Cognition

Individual differences in metacontrast masking regarding sensitivity and response bias

https://doi.org/10.1016/j.concog.2012.04.006Get rights and content

Abstract

In metacontrast masking target visibility is modulated by the time until a masking stimulus appears. The effect of this temporal delay differs across participants in such a way that individual human observers’ performance shows distinguishable types of masking functions which remain largely unchanged for months. Here we examined whether individual differences in masking functions depend on different response criteria in addition to differences in discrimination sensitivity. To this end we reanalyzed previously published data and conducted a new experiment for further data analyses. Our analyses demonstrate that a distinction of masking functions based on the type of masking stimulus is superior to a distinction based on the target–mask congruency. Individually different masking functions are based on individual differences in discrimination sensitivities and in response criteria. Results suggest that individual differences in metacontrast masking result from individually different criterion contents.

Highlights

► Different observers exhibit qualitative different types of masking functions. ► The congruency between target shape and mask shape apparently affects target discrimination. ► These congruency effects reflect response bias rather than discrimination sensitivity. ► Observers differ in discrimination sensitivity and response bias. ► Differences in response bias indicate the use of different perceptual cues.

Introduction

In metacontrast masking the visibility of an otherwise fully visible target stimulus is decreased by a subsequently presented masking stimulus, whose inner contours neatly fit the outer contours of the target stimulus. Under these conditions target visibility is a function of the stimulus onset asynchrony (SOA) between the target and the mask. Two types of masking functions have been traditionally distinguished. With Type A functions target visibility is minimal at short SOAs and increases with increasing SOA. With Type B functions target visibility follows a U-shaped function with minimal target visibility at intermediate SOAs (Kolers, 1962). Although the phenomenon of metacontrast masking has been known and investigated for about a century (e.g. Alpern, 1954, Bachmann, 1984, Bachmann, 1994, Breitmeyer and Ögmen, 2006, Turvey, 1973, Weisstein, 1972, Werner, 1935) and is nowadays widely used as a technique to render stimuli invisible in experimental psychology (e.g., Eimer and Schlaghecken, 1998, Fehrer and Raab, 1962, Klotz and Neumann, 1999, Mattler, 2003, Mattler, 2005, Mattler, 2006, Neumann and Klotz, 1994, Schmidt, 2000, Schmidt, 2002, Vorberg et al., 2003, Vorberg et al., 2004), its underlying mechanisms are not fully understood. In particular, the factors underlying the different types of masking functions remain largely unresolved. Several factors have been isolated in earlier studies, including the ratio of duration and/or intensity of target and mask (Breitmeyer & Ögmen, 2000), the spatial layout of target and mask (Duangudom et al., 2007, Francis and Cho, 2008, Francis and Herzog, 2004), and the instructed task (Breitmeyer et al., 2006). Task dependent masking functions have been explained by the assumption that participants apply different criterion contents when they perform different tasks. Criterion content refers to the stimulus attribute, psychological dimension or perceptual cue on which a judgment is based (Ventura, 1980).

In addition to these factors that influence the type of masking function, we recently have shown that under identical stimulation conditions, masking functions show substantial inter-individual variability. We used a metacontrast masking paradigm in which the target was a small filled square or diamond followed by a mask, whose outer contours was also square or diamond shaped (Fig. 1a and b). Participants had to report the shape of the target stimulus on each trial. Each participant in our study showed either Type-A or Type-B masking. Masking functions developed over the course of two experimental sessions and proved stable over an extended period of up to 23 weeks without further practice. The underpinnings of these individual differences are currently unclear. Individual differences might result from the use of different strategies or informational cues. In addition, individual differences might also be associated with differences in phenomenological experiences (Albrecht et al., 2010, Albrecht and Mattler, 2010, Bachmann, 2010).

In a recent study Maksimov, Murd, and Bachmann (2011) replicated an experiment of our first study (Albrecht et al., 2010). In line with our approach, Maksimov and colleagues found two distinct groups of observers: One group showed Type-A masking functions, and the other group showed “Type-B” masking functions. However, performance in the group with Type-B masking functions decreased only minimally from short to longer SOA. In contrast to our own findings, Maksimov and colleagues employed a different approach for data analysis. Instead of computing d′ separately for each type of mask, these authors distinguished trials according to the relation between the target and the mask in terms of target–mask congruency (Fig. 1b). Authors report that target–mask congruency significantly modulated the effects: Clear Type-A masking occurred in Type-A observes only when target and mask had different outer shapes (incongruent trials). When target and mask had the same outer shape (congruent trials) Type-A observers showed slightly decreasing masking functions. In their “Type B” observers, however, similar masking functions were found on congruent and incongruent trials. Based on this analysis, Maksimov and colleagues conclude that the qualitative differences in masking functions result from incongruent trials only.

However, this conclusion is based on the data of an experiment which failed to find observers with clear Type-B masking functions. As acknowledged by Maksimov and colleagues, performance in their “Type-B” observers was only slightly modulated by SOA and is similar to the performance of Type-A observers on congruent trials. This finding differs from that of our study (Albrecht et al., 2010) in which we found Type-B observers with a clear and steep decrease of performance as a function of SOA. Nonetheless, we thought it is worthwhile to examine the thesis of Maksimov and colleagues by a re-analysis of our data from Albrecht et al. (2010).

Of more importance is, however, that Maksimov and colleagues applied data analyses in a way which differs fundamentally from our analyses. We analyzed participants’ sensitivity in terms of signal detection theory’s d’, which was computed separately for each mask and then averaged across masks. This was done to control for response bias since computing d’ from pooled data across mask types underestimates sensitivity in cases in which response bias varies across mask type (Macmillan & Creelman, 1991). Therefore, the signal detection literature suggests that conditions which may affect response bias should be kept constant when computing sensitivity. This conflicts with computing d’ separately for congruent and incongruent trials because the later analysis necessarily confounds sensitivity with response bias, which leads to inaccurate estimates unless the criterion c = 0 (Vorberg et al., 2004). Note that the same holds for separate assessments of proportion of correct responses on congruent and incongruent trials. To illustrate this complication, imagine an observer who does not comply with the discrimination task but rather responds to the shape of the mask. This observer would be correct on 50% of all trials. This corresponds to chance performance which should be reflected in the data analysis. However, looking only on congruent trials, this observer would be perfectly correct. On incongruent trials, however, he or she would be incorrect on every trial. Therefore, the distinction of congruent and incongruent trials would lead to large performance differences although the true sensitivity is zero in all conditions. In sum, it is unclear whether the data pattern reported by Maksimov et al. (2011) reflects differences in sensitivity or differences in response bias.

Here we present a re-analysis of the data of Albrecht et al. (2010), Experiment 1, with congruency as additional variable and an analysis of the data of a new experiment which was conducted to investigate whether individuals differ in terms of response biases. To anticipate our results, our data suggest that performance in Type-A observers differs on congruent and incongruent trials at short SOAs, which is in line with the findings reported by Maksimov and colleagues (2011). In contrast to their findings, however, our data shows that Type-B observers exhibit strong Type-B masking on both congruent and incongruent trials. Most important, however, the data of both experiments strongly suggests that the congruence effects for Type-A observers can be reduced to their use of information provided by the mask, and, thus, reflect differences in response bias rather than sensitivity.

Section snippets

Experiment 1

First, we reanalyzed the data reported in Experiment 1 of Albrecht et al. (2010) with target–mask congruence as additional factor despite the above mentioned methodological issues. In particular it is of interest whether differences between groups of observers are limited to differences on incongruent trials. We hypothesize that the similarity for Type-A and Type-B observers on congruent trials in Maksimov et al. (2011) comes from their failure to find pronounced Type-B masking. In contrast, in

Experiment 2

As hypothesized in our original report (Albrecht and Mattler, 2010, Albrecht et al., 2010) and also by Bachmann and colleagues (Bachmann, 2010, Maksimov et al., 2011), both types of observers may rely on different perceptual experiences of the target–mask sequence to solve the task. Phenomenological reports on a trial-by-trial basis suggest that Type-A observers spontaneously perceive and use apparent motion in the target–mask sequence as a cue, whereas Type-B observers perceive a kind of

General discussion

Both, the re-analysis of our earlier study and a new experiment using slightly different parameters for presentation, consistently show that differences between Type-A and Type-B observers are not limited to incongruent trials. Similar to Maksimov and colleagues (2011) Type-A observers showed Type-A masking on incongruent trails and no clear effect of SOA on congruent trials. But in contrast to Maksimov et al., Type-B observers showed identical and pronounced Type-B masking functions on both,

References (41)

  • T. Bachmann

    The process of perceptual retouch: nonspecific afferent activation dynamics in explaining visual masking

    Perception and Psychophysics

    (1984)
  • T. Bachmann

    Psychophysiology of visual masking: The fine structure of conscious experience

    (1994)
  • B.G. Breitmeyer et al.

    Recent models and findings in visual backward masking: a comparison, review, and update

    Perception and Psychophysics

    (2000)
  • B.G. Breitmeyer et al.

    Visual masking: Time slices through conscious and unconscious vision

    (2006)
  • B. Bridgeman et al.

    Interaction of stimulus size and retinal eccentricity in metacontrast masking

    Journal of Experimental Psychology: Human Perception and Performance

    (1979)
  • M. Bruchmann et al.

    Metacontrast masking within and between visual channels: Effects of orientation and spatial frequency contrasts

    Journal of Vision

    (2010)
  • V. Di Lollo et al.

    Stimulus-onset asynchrony is not necessary for motion perception or metacontrast masking

    Psychological Science

    (1993)
  • V. Di Lollo et al.

    Competition for consciousness among visual events: The psychophysics of reentrant visual processes

    Journal of Experimental Psychology: General

    (2000)
  • V. Duangudom et al.

    What is the strength of a mask in visual metacontrast masking?

    Journal of Vision

    (2007)
  • M. Eimer et al.

    Effects of masked stimuli on motor activation: Behavioral and electrophysiological evidence

    Journal of Experimental Psychology: Human Perception and Performance

    (1998)
  • Cited by (0)

    View full text