Elsevier

Consciousness and Cognition

Volume 17, Issue 3, September 2008, Pages 646-656
Consciousness and Cognition

Undetected changes in visible stimuli influence subsequent decisions

https://doi.org/10.1016/j.concog.2007.03.002Get rights and content

Abstract

Change blindness—our inability to detect changes in a stimulus—occurs even when the change takes place gradually, without any disruption [Simons, D. J., Franconeri, S. L., & Reimer, R. L. (2000). Change blindness in the absence of a visual disruption. Perception, 29(10), 1143–1154]. Such gradual changes are more difficult to detect than changes that involve a disruption. Using this method, David et al. [David, E., Laloyaux, C., Devue, C., & Cleeremans, A. (in press). Change blindness to gradual changes in facial expressions. Psychologica Belgica] recently showed substantial blindness to changes that involve facial expressions of emotion. In this experiment, we show that people who failed to detect any change in the displays were (1) nevertheless influenced by the changing information in subsequent recognition decisions about which facial expression they had seen, and (2) that their confidence in their decisions was lower after exposure to changing vs. static displays. The findings therefore support the notion that undetected changes that occur in highly salient stimuli may be causally efficacious and influence subsequent behavior. Implications concerning the nature of the representations associated with undetected changes are discussed.

Introduction

Change blindness, our inability to detect large changes in visual displays, is a striking phenomenon that has now been demonstrated through various paradigms (e.g. Rensink, 2002, Rensink et al., 2000, Rensink et al., 1997, Simons, 2000, Simons and Levin, 1998, Simons and Levin, 2003).

While early empirical work in this domain was characterized by the use of highly artificial stimuli consisting of dot matrices (Phillips, 1974) or letter arrays (Pashler, 1988), more recent demonstrations involve changes that occur in complex, realistic scenes (Rensink et al., 1997, Rensink et al., 2000, Simons et al., 2000, Velichkovsky et al., 2002). Change blindness challenges introspective judgments that our perception of the world is complete and accurate—a belief so strongly held that it has been dubbed “change blindness blindness” (Levin, Momen, Drivdahl, & Simons, 2000). For instance, in real life situations, observers fail to notice changes as dramatic as switching an individual with whom they are interacting to a different person (Levin & Simons, 1997). Moreover, observers continue to exhibit change blindness even when directly instructed to detect changes (Rensink et al., 1997, Rensink et al., 2000, Simons et al., 2000). Numerous relevant studies have used the so-called “flicker” paradigm, in which two images, identical to each other but for a single change, are displayed alternatively for 240 ms and separated from each other by a 80 ms blank screen—the disruption. This flickering sequence is typically repeated until the observer detects the change. It can take many such cycles (up to hundreds in some conditions) for observers to become aware of the change (Rensink et al., 1997). Rensink et al. (1997) have documented that detection rate is influenced by the location at which the change occurs. For instance, performance improves when the change involves an area of major interest compared to an area of marginal interest.

In such flicker studies (Rensink et al., 1997, Rensink et al., 2000, Simons et al., 2000), change blindness is induced by a brief visual disruption inserted between the two target images, and hence the demonstration is perhaps less compelling than it might first appear because the images only appear for a brief duration. “Gradual change” studies address this limitation by introducing changes only very progressively (e.g., over a period of 12 s), without any disruption. With this method, observers have the opportunity to look at the image carefully2 and in an uninterrupted manner. Studies using this paradigm have explored people’s ability to detect changes in object color or changes involving the deletion or addition of an object. Simons et al. (2000) showed that gradual change produces powerful change blindness despite the fact that the change is continuously happening in front of the observer. Moreever, change blindness rate was lower under gradual change conditions than under disruption conditions for changes involving deletion or addition. However, as Simons et al. (2000) pointed out, this result might be due to artifacts. Indeed, gradual changes, when they involve deletion or addition, necessarily produce intermediate frames that contain easily noticed artifacts, such as transparent objects (Simons et al., 2000). In a second experiment that avoided such artifacts by using color changes, Simons et al. (2000) showed that change blindness rate was higher for gradual changes than for disruption changes (see also Auvray & O’Regan, 2003 for similar conclusions).

In this light, David, Laloyaux, Devue, and Cleeremans (in press)3 investigated gradual changes with facial expressions of emotions because expression changes involve essentially modifications to the shape and spatial relationships of features that are internal to an object rather than the appearance or disappearance of new objects. They showed that such changes still produce quite substantial rates of change blindness, as only about 15% of participants detected slow, gradual changes in the facial expression of actors, even when such changes occur in front of their eyes as they intentionally scrutinize the stimuli under direct instructions to detect any changes.

These and other findings leave open an important question, however: what is represented, if anything, when a change remains undetected? Authors have suggested several explanations (see, Simons, 2000, for a review). A first possibility is that there is a failure to encode or represent the pre-change and/or post-change information (O’Regan & Noe, 2001). This absence of visual representation would explain change blindness because comparison with the post-change information is then be impossible. Another possibility is that the representation of the pre-change information is erased or overwritten by the post-change information (Beck and Levin, 2003, Becker et al., 2000, Landman et al., 2003, Tatler, 2001). Other authors have suggested that it is also possible that while information about both pre- and post-change states is stored in some manner, the comparison between the two representations fails, which would likewise produce change blindness (Angelone et al., 2003, Hollingworth and Henderson, 2002, Mitroff et al., 2004, Scott-Brown et al., 2000; Silverman and Mack, 2006, Simons et al., 2002). In this latter case, the problem is thus not about representations, but rather about an operation that has to be performed on them.

Simons (2000) usefully summarized five alternatives as follows: (1) “Overwriting”, which means that pre-change representation would be erased by post-change representation (Beck and Levin, 2003, Becker et al., 2000, Landman et al., 2003, Tatler, 2001), (2) “First impression”, which suggests that it would only be the pre-change representation that is represented and not what is presented after the change. This account makes it difficult, however, to explain why the visual system would stop forming representations when a change is displayed, as representations are updated with time (Hollingworth & Henderson, 2004), (3) “Nothing is stored” is a position defended by O’Regan and Noe (2001), and which posits that no visual representation should be postulated to perceive the world. (4) “Everything is stored but nothing is compared” assumes that some representation is formed both for the pre-change and the post-change display but that no comparison is made between both representations when change blindness occurs. As noted earlier, much recent evidence suggests that this possibility might explain some change blindness cases (Angelone et al., 2003, Hollingworth and Henderson, 2002, Mitroff et al., 2004, Scott-Brown et al., 2000, Silverman and Mack, 2006, Simons et al., 2002). Finally, (5) “Feature combination” suggests that an integrated representation would be formed based on both pre- and post-change displays, which would then make change detection impossible.

Leaving aside for the time being the question of which of these accounts is most likely to be correct, the question of whether undetected changes are represented somehow was directly addressed in a convincing study by Mitroff et al. (2004), who showed that observers remain capable of recognizing, in a forced-choice task, both the pre- and post-change information, even when reporting being unaware of having detected a change. In that study, observers were exposed to changes in a classic change detection paradigm (Phillips, 1974) in which 4–8 objects were displayed on a computer screen. On most trials, one of the objects changed into another object. Right after the change detection trial, a series of subsequent two-alternative forced-choice (2AFC) trials was administered in random order. One 2AFC trial presented the pre-change object along with a random novel object that had not been presented in the trial, and participants were asked to choose which had been presented. Another 2AFC trial presented the post-change object along with a random novel object. A third 2AFC trial presented a random presented non-changing object along with a novel object. Observers were asked to report if they had detected the change after (Experiments 1–3) or before (Experiment 4) the series of 2AFC trials. Through this design, Mitroff et al. showed that even when subjects were unaware of the change, they were nevertheless able to remember, better than expected by chance, both the pre- and post-change information.

However, the question of determining what representation observers form of the material when exposed to changing stimuli that contain more than two images has never been explored directly. When only two images (i.e., the pre- and the post-change displays) are presented, as in the flicker paradigm, subjects may remember the pre-change display, the post-change display, or both, as shown by Mitroff et al. But what is remembered when many intermediate images are presented, as in the gradual change paradigm? Do participants remember only the first or the last display? Or do they construct some average representation of the entire sequence? These are essentially the questions we explored in this study.

Instead of testing every aspect of the representation observers have formed of a changing stimulus, which is impractical, we asked what stimulus subjects spontaneously select when asked to identify what they had seen among a limited number (i.e., five) of perceivably different images. Crucially, in our paradigm, participants were not informed that a change may occur, and we focused data analysis on those participants who had remained unaware of the fact that some stimuli were changing. This is important for previous studies (Angelone et al., 2003, Hollingworth and Henderson, 2002, Mitroff et al., 2004, Scott-Brown et al., 2000, Silverman and Mack, 2006, Simons et al., 2002) have demonstrated that some information both about pre- and post-change displays is stored when observers are forewarned that a change may occur.

This issue is also related to recent debates about implicit change detection (Fernandez-Duque et al., 2003, Fernandez-Duque and Thornton, 2000, Fernandez-Duque and Thornton, 2003, Laloyaux et al., 2006, Mitroff et al., 2002, Thornton and Fernandez-Duque, 2000, Thornton and Fernandez-Duque, 2002). For instance, Fernandez-Duque and Thornton (2000) presented their subjects with changes of orientation of horizontal and vertical rectangles and showed that, even when subjects reported being unaware of a change, they were nevertheless able to localize the change above chance level. In a different paradigm, Thornton and Fernandez-Duque (2000) again presented their subjects with horizontal and vertical rectangles but showed that exposure to a change in the orientation of an item produced a congruency effect in a subsequent judgment task about the orientation of a rectangle. Although these studies were criticized by Mitroff et al. (2002), it seems that these effects persist even after the correction of potential biases (Fernandez-Duque and Thornton, 2003, Laloyaux et al., 2006). It therefore seems that a change that observers report as being unaware of can influence subsequent behavior.

This question, however, remains open for complex stimuli that involve more than two subsequent brief displays of rectangles. A recent study (Hollingworth & Henderson, 2004) is relevant in this context. The authors used what could be called a “gradual flicker paradigm”, wherein a scene was progressively rotated in steps of 1° of visual angle on each display, with each display separated from the next one by a brief blank screen. The results indicated that half of the participants were unaware of the change up to a rotation of 48°. However, when a shift (using a blank to avoid low level signal) back to the original scene was produced after a cumulated rotation of 20° or 30° that had remained undetected, most subjects could then clearly see the change, suggesting the existence that visual memory had been updated unconsciously. This conclusion follows from the fact that had visual memory not been updated at all, subjects would not have been able to detect the change when coming back to the original display.

In the present study, we thus explored the same issue, but using a fully continuous stimulus, and a different reporting methodology that allows for somewhat stronger conclusions. Specifically, in this experiment, we used the subset of stimuli involving gradual changes of facial expressions used by David et al. (in press) to investigate what representation observers form when exposed to a change while remaining unaware of this change. If nothing at all is represented, then one should expect to observe a completely flat distribution of recognition choices, that is, observers should choose all of the intermediate images equally often. It should be noted that the same prediction can be made if “everything is represented”. However, if the distribution of choices differs from pure chance, (e.g., if it exhibits a primacy or recency effect, or if some other bias towards particular intermediates is apparent), this would suggest unconscious sensitivity to change, or at least an update of the representation during the change. It is important to realize that finding a distribution of choices that differs from a flat distribution rules out both the “nothing is represented” hypothesis and the “everything is equally represented” hypothesis.

Note that the task is prima facie identical for static and for changing stimuli, that is, it is just as easy for people to express their choice after having been exposed to a changing or to a static stimulus, given that we only considered participants who had remained unaware of the changes. In other words, all participants whose results we analyze below think that they had been exposed to static, photographic stimuli.

To summarize, in this experiment, we used the subset of stimuli involving gradual changes of facial expressions used by David et al. (in press) to investigate what representation observers form when exposed to a change while remaining unaware of this change.

Section snippets

Subjects

Forty-nine undergraduate students from the Université Libre de Bruxelles participated either for partial class credit or for four Euros.

Materials

The material was composed of the eight stimuli involving facial expression changes used in David et al. (in press). David et al. used eight different scenes composed of three actors (three different actors for each scene) showing various facial expressions and placed in different locations. The scenes were either static, or they contained a facial expression

Results

Fourteen out of 49 participants reported awareness of at least one change. They were thus considered “aware” and discarded from the analysis, which thus concerned “unaware” participants only. Thus, 35 out of 49 (71.4%) participants failed to detect any change and were included in the analysis.

We examined performance on the recognition task by considering the distribution of choices as a function of whether the stimulus had been changing or not (see Fig. 3). While there was a single correct

Discussion

To assess what is represented when one is confronted with an undetected change, participants were exposed either to static or gradually changing pictures of faces for 12 s and asked to memorize them. Changes were applied to the emotional expression displayed by one of the actors, going from neutral to emotional for half of the stimuli and in the reverse direction for the other half. In a subsequent recognition test that followed each trial, participants were then shown five intermediate morphed

Acknowledgments

C.L. and C.D. are Research Fellows with the National Fund for Scientific Research (Belgium). A.C. is a research director with the same institution. This work was supported by an institutional grant from the Université Libre de Bruxelles and by Concerted Research Action 06/11-342 titled “Culturally modified organisms: What it means to be human in the age of culture”, financed by the Ministère de la Communauté Française—Direction Générale de l’Enseignement non obligatoire et de la Recherche

References (38)

  • David, E., Laloyaux, C., Devue, C., & Cleeremans, A. (in press). Change blindness to gradual changes in facial...
  • P. Ekman et al.

    Constants across cultures in the face and emotion

    Journal of Personality and Social Psychology

    (1971)
  • D. Fernandez-Duque et al.

    Representation of change: Separate electrophysiological markers of attention, awareness, and implicit processing

    Journal of Cognitive Neuroscience

    (2003)
  • D. Fernandez-Duque et al.

    Change detection without awareness: Do explicit reports underestimate the representation of change in the visual system?

    Visual Cognition

    (2000)
  • D. Fernandez-Duque et al.

    Explicit mechanisms do not account for implicit localization and identitication of change: An empirical reply to Mitroff et al. (2002)

    Journal of Experimental Psychology-Human Perception and Performance

    (2003)
  • A. Hollingworth

    Failures of retrieval and comparison constrain change detection in natural scenes

    Journal of Experimental Psychology-Human Perception and Performance

    (2003)
  • A. Hollingworth

    Constructing visual representations of natural scenes: The roles of short- and long-term visual memory

    Journal of Experimental Psychology-Human Perception and Performance

    (2004)
  • A. Hollingworth et al.

    Accurate visual memory for previously attended objects in natural scenes

    Journal of Experimental Psychology-Human Perception and Performance

    (2002)
  • A. Hollingworth et al.

    Sustained change blindness to incremental scene rotation: A dissociation between explicit change detection and visual memory

    Perception & Psychophysics

    (2004)
  • Cited by (12)

    • Change Blindness

      2015, International Encyclopedia of the Social & Behavioral Sciences: Second Edition
    • Priming effects under correct change detection and change blindness

      2013, Consciousness and Cognition
      Citation Excerpt :

      In the present research, we will examine these questions by considering the possible effects of an unperceived change on an immediately successive semantic categorization task. Priming effects under CB are well documented when an image change is followed by a perceptual task (Fernandez-Duque & Thornton, 2000; Fernandez-Duque & Thornton, 2003; Laloyaux, Devue, Doyena, David, & Cleeremans, 2008; Laloyaux, Destrebecqz, & Cleeremans, 2006; Thornton & Fernandez-Duque, 2000). What has been less studied is whether CB can elicit priming effects on performance in a semantic task performed immediately after the image change (e.g., word-picture matching, picture categorization, word categorization).2

    • Action blindness in response to gradual changes

      2010, Consciousness and Cognition
    View all citing articles on Scopus
    1

    Fax: +32 366 28 59.

    View full text