Change blindness blindness: Beliefs about the roles of intention and scene complexity in change detection

https://doi.org/10.1016/j.concog.2006.01.003Get rights and content

Abstract

Observers have difficulty detecting visual changes. However, they are unaware of this inability, suggesting that people do not have an accurate understanding of visual processes. We explored whether this error is related to participants’ beliefs about the roles of intention and scene complexity in detecting changes. In Experiment 1 participants had a higher failure rate for detecting changes in an incidental change detection task than an intentional change detection task. This effect of intention was greatest for complex scenes. However, participants predicted equal levels of change detection for both types of changes across scene complexity. In Experiment 2, emphasizing the differences between intentional and incidental tasks allowed participants to make predictions that were less inaccurate. In Experiment 3, using more sensitive measures and accounting for individual differences did not further improve predictions. These findings suggest that adults do not fully understand the role of intention and scene complexity in change detection.

Introduction

Suppose you were approached as a potential witness to a purse theft. At the time of the theft, you were sitting in the waiting room of a doctor’s office reading a magazine. Just prior to the proposed theft, you looked up from your magazine at the scene in front of you for a few seconds and then, just as a man is walking through the room, you look back down at your magazine briefly. While you are looking at the magazine, unbeknown to you, the man walking through the room picks up a purse that was sitting on a chair in front of you. Then you look back up at the room in front of you and a nurse walks into the waiting room and asks you if the man took anything from the room. Having an accurate understanding of your ability to detect changes in the visual environment is important to your ability to judge whether you will be an accurate witness for the crime. For example, if you believe you would be able to accurately detect the disappearance of an object, and you did not detect a change, you may incorrectly testify that the man could not have taken anything. Here, we examine what factors (e.g., mental effort and scene complexity) affect people’s beliefs about their ability to notice visual changes such as the disappearance of objects from a previously viewed scene.

Integral to our ability to detect the disappearance of an object from a visual scene may be our intent to remember objects in our visual field. Recent research has demonstrated the importance of effort in vision by showing that participants fail to notice large changes to objects in complex natural scenes and artificial displays (change blindness; Blackmore et al., 1995, Grimes, 1996, Henderson, 1997, Levin and Simons, 1997, McConkie and Currie, 1996, O’Regan et al., 1999, Pashler, 1988, Phillips, 1974, Rensink, 2000, Rensink et al., 1997, Simons, 1996, Simons and Levin, 1998; for a review see Simons, 2000). Change blindness is robust across numerous change detection paradigms (see Simons and Rensink, 2005a, Simons and Rensink, 2005b for review), but it contrasts strongly with observers’ predictions that they will be able to detect most visual changes (change blindness blindness; Levin et al., 2002, Levin et al., 2000). This suggests that observers may have an inaccurate belief that an abundance of visual information is automatically stored in memory and available for retrieval over time. Here, we examine the importance of actively searching for changes (intentional change detection) in reducing change blindness (CB) and the degree to which observers are aware of the impact of intentionally directing metal effort on change detection performance.

Change blindness results from a failure of visual awareness and these failures are more prevalent than many would predict given our phenomenological experience of being visually aware of the world around us (see Varakin, Levin, & Fidler, 2004 for review). For example, a failure to detect a visual change resulted in the loss of many lives in the crash of a French Airbus AT320-111 near Strasbourg, France in 1992. This crash has been attributed to the pilot’s failure to notice a mode-signal change presented directly in sight on the aircraft’s flight control computer (Varakin et al., 2004). In addition, continuity errors occur in films very frequently (e.g., there is a coffee cup in one scene and it is gone in the next scene). There are several web sites devoted to the art of detecting these discontinuities (e.g., http://www.moviemistakes.com/ and http://www.jonhs.com/moviegoofs/). Given that changes are prevalent in everyday visual experiences, why do people remain so poor at detecting them?

When a visual change occurs, it is accompanied by an abrupt onset or motion transient that can attract attention to the change and increase the likelihood that the viewer will be aware of the change (for an overview see Rensink, 2002). However, if the abrupt onset or motion transient occurs outside of the viewer’s field of view (e.g., the viewer is looking away from the changing object, another object is temporarily occluding the viewer’s view of the object, or the change happens while the viewer is making an eye movement), attention will not automatically be directed to the location of the change and the change is far less likely to be detected (change blindness). For example, when driving down the road, it is important to monitor the state of the car in front of us. If, while a driver is looking down to adjust the radio, the break lights of the preceding car turn on, the driver will be less likely to notice the change than if they were looking in the direction of the lights at the time of the change. Therefore, change blindness can occur frequently in the real world because in the absence of perceivable motion transients, limited capacity processes such as attention and memory are necessary for change detection to occur (Levin and Simons, 1997, Simons, 2000, Simons and Rensink, 2005a, Simons and Rensink, 2005b).

In the absence of bottom-up cues (abrupt on-sets and motion transients) to direct attention to the location of the change, the visual system must rely on top-down processes to allocate attention to the location of potential changes. For example, using a change detection task in which the motion transient accompanying each change was masked, Beck, Angelone, and Levin (2004) demonstrated that knowledge about which changes were likely to occur modulated change detection performance. Specifically, participants were more likely to detect probable changes (e.g., a lamp turning from off to on) than improbable changes (e.g., a blue lamp changing into an green lamp; Beck et al., 2004). This occurs because processing resources are preferentially allocated to aspects of the visual world that are likely to change, thereby increasing the likelihood that these changes will be detected (Beck, Peterson, & Angelone, in press). Therefore, successful change detection is often dependant on the top-down direction of cognitive processes to the location of the change. When participants are aware that changes are going to occur (an intentional change detection task), top-down allocation of processing resources may be more likely to occur than in a situation in which participants are not expecting changes to occur (an incidental change detection task; see Ackerman, 1985 for a discussion of the difference between intentional and incidental tasks). Therefore, it is expected that change detection performance in an intentional change detection task will be better than in an incidental change detection task (the CB intention hypothesis).

Exploring the role of intent in change detection is an important question for elucidating the various impacts of memory on change detection. Accurate visual memory representations of pre-change stimuli must be encoded and maintained in memory for change detection to occur (see Simons, 2000 for review). This process may occur automatically as proposed by visual memory theory (Hollingworth and Henderson, 2002, Hollingworth et al., 2001) or intentional encoding and maintenance may be necessary for successful change detection. If intentional encoding and maintenance improves change detection performance, it may do so through maximizing the use of capacity limited memory systems such as short-term memory (STM). Research has demonstrated that change detection performance declines as the number of items in a display increases because of the employment of limited capacity attentional and memory systems (Beck and Levin, 2003, Wright et al., 2000, Zelinsky, 2001). Therefore, if intention improves change detection performance by maximizing the use of these capacity limited processes, this strategy should become more effective as the number of objects in the scene (scene complexity) increases. That is, the difference between intentional and incidental change detection performance should increase as scene complexity increases (the CB scene complexity hypothesis).

When participants are asked to predict their ability to detect visual changes, they consistently predict that they will be able to detect changes that are often not detected (change blindness blindness; Beck et al., 2004, Levin et al., 2002, Levin et al., 2000, Scholl et al., 2004). Referring back to the scenario presented at the beginning of this paper, a viewer’s beliefs about their ability to detect the disappearance of objects from their visual world could be directly relevant to the viewer’s ability to be a credible witness to a crime. For example, the viewer may have failed to detect the disappearance of the purse (experiencing CB), and believe that if the purse were there they would have detected its disappearance. In this case the viewer would testify that they are confident that the purse was not in the chair before the man walked through the waiting room. Alternatively, the viewer may have failed to detect the disappearance of the purse (experiencing CB), but be aware that their ability to detect these types of changes may be low. In this case, the viewer would testify that they are not confidant that the purse was not in the chair before the man walked through the waiting room. Viewers may be more likely to be overconfident witnesses because they are unaware of what types of factors will affect their ability to detect visual changes.

Although CB suggests that vision is a limited capacity process, participants may have the feeling that they have unlimited access to visual information because vision seems effortless and as though there is immediate access to everything in the external world (Dennett, 1991, Gibson, 1979, O’Regan, 1992). Therefore, participants may not be aware of the link between intentionally guiding attention and our ability to monitor the visual world over time. For example, even though probable changes are detected more frequently than improbable changes, participants predict equal levels of CB for both types of changes. This suggests that the process of allocating visual processing resources based on top-down knowledge is not a deliberate or conscious process (Beck et al., 2004). However, it is unclear as to whether participants were unaware that processing resources are directed preferentially toward probable changes or if they have a more general lack of awareness for the role of allocating processing resources in improving their ability to monitor objects for change over time. Here, we explore the possibility that observers fail to understand the extent to which directing processing resources can improve performance on an attention demanding task such as change detection (the CBB intention hypothesis).

Change blindness blindness (CBB) appears to occur due to a general lack of awareness of the processes involved in change detection. Levin et al. (2000) demonstrated CBB by showing that participants predicted they would be 83% accurate in detecting changes that only 11% of participants actually detected in the CB studies (Levin and Simons, 1997, Simons and Levin, 1998). This finding persisted even when participants were asked to predict performance for a situation when the pre- and post-change shots were separated by over an hour. Furthermore, in open-ended response justification questions, less than 15% of participants mentioned memory as an important factor in CBB, some participants commented that changes would “just pop out,” and many indicated that attention to the changing object was not necessary for detecting changes (Levin et al., 2002). This suggests that participants lack an understanding of the roles of limited capacity cognitive processes (e.g., memory and attention) in change detection suggesting that the activation of these processes may be largely unavailable to conscious inspection. Therefore, participants are likely to be unaware of the effects of scene complexity on overloading these capacity limited processes and that this overloading can be minimized by directing attention to the change detection task (CBB scene complexity hypothesis).

Section snippets

Current experiments

Experiment 1 examined the difference between performance on intentional and incidental change detection tasks. Both tasks involved presentation a pre-change scene and then after a brief disruption, presentation of a post-change scene in which one of the objects was replaced with another object. In the intentional task participants were told that changes would occur in the scenes and that their task was to detect the changes. In the incidental task, participants were told to search the scene for

Participants

One hundred seventy five introductory psychology students at Kent State University participated in exchange for class credit. Each participant completed one of the four conditions: 23 completed the intentional performance condition, 102 completed the incidental performance condition, 25 completed the intentional prediction condition, and 25 completed the incidental prediction condition. There were more participants in the incidental performance condition because each participant in this

Experiment 2

In Experiment 2, we further investigated the degree to which participants are aware of the effects of intention and scene complexity on CBB. We attempted to make knowledge about performing incidental tasks more accessible by providing explicit reference to the divided attention nature of the incidental change detection task and by allowing participants to compare intentional and incidental change detection tasks. Specifically, we explored the possibility that participants will more accurately

Experiment 3

In Experiment 3, we further tested the impact of intention and scene complexity on CBB by examining possible individual differences in beliefs about the role of intention and by using more sensitive prediction measures. To examine possible individual differences in knowledge about intention and change detection, we explicitly asked participants whether they would need to be purposefully looking for the changes to detect them. Research has shown that children who score high on a perceptual

General discussion

In Experiment 1, participants predicted almost equal change detection performance for both incidental and intentional tasks (only a 3% difference). However, participants were significantly better at detecting changes in an intentional task than an incidental task (a 40-53% difference). Therefore, when predicting change detection performance, participants did not have a readily accessible understanding about the role of intention in detecting changes. In Experiment 2, when participants were told

References (44)

  • S.J. Blackmore et al.

    Is the richness of our visual world an illusion? Transsacadic memory for complex scenes

    Perception

    (1995)
  • D.C. Dennett

    Consciousness explained

    (1991)
  • R.A. Diana et al.

    Visual versus verbal metacognition: Are they really different

  • B. Fischhoff

    Hindsight is not equal to foresight: The effect of outcome knowledge on judgment under uncertainty

    Journal of Experimental Psychology: Human Perception and Performance

    (1975)
  • C.E. Granrud

    Visual metacognition and the development of size constancy

  • J.J. Gibson

    The ecological approach to visual perception

    (1979)
  • A. Gopnik et al.

    The theory

  • J. Grimes

    On the failure to detect changes in scenes across saccades

  • E.M. Harley et al.

    The “saw-it-all-along” effect: Demonstrations of visual hindsight bias

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (2004)
  • J.M. Henderson

    Transsacadic memory and integration during real-world object perception

    Psychological Science

    (1997)
  • 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.

    To see and remember: Visually specific information is retained in memory from previously attended objects in natural scenes

    Psychonomic Bulletin & Review

    (2001)
  • Cited by (41)

    • Metacognitive judgements of change detection predict change blindness

      2022, Cognition
      Citation Excerpt :

      These two features make change blindness easy to demonstrate across different experimental paradigms (Levin & Simons, 1997; Levin & Simons, 2000; O'Regan et al., 1999; Rensink et al., 1997; Simons & Rensink, 2005) and in real world settings (e.g., Simons & Levin, 1998). Despite the magnitude (in size and duration) of change blindness, people are generally unaware that they experience change blindness at all, resulting in “change blindness blindness” (Beck, Angelone, & Levin, 2004; Beck, Levin, & Angelone, 2007; Levin, Drivdahl, Momen, & Beck, 2002; Levin, Momen, Drivdahl, & Simons, 2000; Scholl, Simons, & Levin, 2004). Change blindness blindness is robust and has been shown to occur for a range of durations in between changes (e.g., one minute or one hour), for varying number of objects in the scene (i.e., scene complexity), and even when participants are reminded about the role of memory and attention in change detection (Beck et al., 2007; Levin et al., 2002).

    • What can magic reveal about the brain

      2021, Encyclopedia of Behavioral Neuroscience: Second Edition
    • Change detection in urban and rural driving scenes: Effects of target type and safety relevance on change blindness

      2017, Accident Analysis and Prevention
      Citation Excerpt :

      A final point worth noting is that the self-report measures of cognitive failures and driving-related errors and lapses did not reliably predict change detection performance. This is reminiscent of “change blindness blindness”, whereby observers under-estimate their susceptibility to change blindness (Beck et al., 2007). When driving, this could be problematic if drivers are not aware of precisely how difficult it is to detect changes, especially for changes involving road signs.

    • Use of change blindness to measure different abilities to detect relevant changes in natural driving scenes

      2012, Transportation Research Part F: Traffic Psychology and Behaviour
      Citation Excerpt :

      Thus, change blindness may enhance differences in the expectation and relevance of dangers according to the level of driving experience. Concerning the ability to detect changes according to the complexity of the scene, many studies have demonstrated that change detection performance declines as the number of items in a display increases (Beck & Levin, 2003; Beck, Levin, & Angelone, 2007; Wright, Green, & Baker, 2000). Thus, change detection may show differences in the capacity to focus on change depending on the complexity of the road setting.

    • Exploring the informational sources of metaperception: The case of Change Blindness Blindness

      2011, Consciousness and Cognition
      Citation Excerpt :

      There are, however, two apparently contradictory ways of using the local effort cue in order to assess one’s performance in a change detection task. Beck, Levin, and Angelone (2007) found that perceivers tend to reason that more effort allocation in a change detection task is predictive of better performance. On the other hand, our study shows that a larger amount of effort in a given trial is used as a diagnostic cue of poorer performance (in agreement with Borg & Noble, 1974; Petrusic & Baranski, 1989).

    View all citing articles on Scopus

    This material is based upon work supported by the National Science Foundation under Grant No. 0214969 to DTL.

    View full text