When you fail to see what you were told to look for: Inattentional blindness and task instructions
Highlights
► Four black squares and four white diamonds in a dynamic selective-looking paradigm. ► Observers instructed to count bounces by squares did not notice an unexpected square. ► Observers instructed to count bounces by diamonds did notice the unexpected square. ► Task instructions specified shape but observers developed an attentional set for the associated colour. ► Attention was set for colour because this allowed easier discrimination of task-relevant from task-irrelevant items.
Introduction
Individuals can set their attention to be differentially sensitive to specific properties of the visual world. This attentional readiness facilitates selective processing and underlies task efficiency (Folk et al., 1992, Folk et al., 1994), but it does so at a cost. Once an attentional set is adopted, an unexpected object or event may go undetected if it does not share set properties, such as colour (Koivisto and Revonsuo, 2008, Most et al., 2001, Simons and Chabris, 1999), shape (Most, Scholl, Clifford, & Simons, 2005), luminance (Most et al., 2005), category (Koivisto et al., 2004, Most, in press), number (White & Aimola Davies, 2008), or semantic meaning (Koivisto and Revonsuo, 2007, Koivisto and Revonsuo, 2009). The striking failure to notice an unexpected object or event in plain sight when attention is otherwise engaged is referred to as inattentional blindness (Mack & Rock, 1998).
The best-known example of inattentional blindness comes from the gorilla study that was made famous by Simons and Chabris (1999; see Neisser & Becklen, 1975 who inspired the paradigm, and Simons, 2010 for a recent adaptation of the paradigm). Observers viewed a brief video of two teams, one dressed in white shirts and the other in black. Observers were told “that they would be watching two teams of three players passing basketballs and that they should pay attention to either the team in white (the White Condition) or the team in black (the Black Condition)” (p. 1066). The task was to count the number of passes made by the specified team. In one version of the video, a person dressed in a black gorilla suit walked through the centre of the action, from the left to the right side of the screen, remaining visible for 5 s. Incredibly, the gorilla was reported by only 42% of observers in the White Condition compared with 83% of observers in the Black Condition (results for opaque video, easy task). One explanation for the significantly better performance by observers counting the passes made by the team in black is that the colour of the unexpected gorilla matched the observers’ attentional set for black.
Most et al., 2001, Most et al., 2005 have rigorously tested the role of attentional set in inattentional blindness with dynamic computer-generated stimuli rather than with the naturalistic videos used in the pioneering studies. In the paradigm most relevant to the current study, Most et al. (2005, Experiment 1) investigated attentional set for colour (black or white) or for shape (circle or square). Stimuli comprised two black circles, two black squares, two white circles, and two white squares. These eight items moved around the computer display independently, with varying trajectories and speeds, and periodically bounced off the edges of the display. The task was to count the total number of times that the items in a specified subset bounced off the edges of the display. For example, observers were instructed to count bounces by items that were (1) black (circles and squares), (2) white (circles and squares), (3) circles (black and white), or (4) squares (black and white). Five seconds into the third trial, an additional black circle moved from right to left across the horizontal extent of the display, remaining visible for 5 s. This critical stimulus was reported by 88% of observers counting bounces by black items and 81% of observers counting bounces by circles. For these two groups of observers, the critical stimulus shared the property specified in their task instructions – black or circle. In contrast, the critical stimulus was reported by 0% of observers counting bounces by white items and 6% of observers counting bounces by squares. For these two groups of observers, the unexpected black circle did not share the property specified in their task instructions – white or square.
An important feature of this series of experiments by Most et al. (2005) was that shape and colour varied orthogonally. Colour could not be used to discriminate between circles and squares; and shape could not be used to discriminate between black items and white items. Thus, the property that was explicitly specified in the task instructions (e.g., count bounces by circles) was the only property that could be used to discriminate between task-relevant items (e.g., black circles and white circles) and task-irrelevant items (e.g., black squares and white squares). Consequently, it is reasonable to assume that the observers adopted an attentional set for the property specified in the task instructions and maintained that set until the critical stimulus was presented. Indeed, the assumption that an attentional set corresponding to the task instructions is maintained until the critical stimulus is presented is, quite properly, typical of inattentional blindness research.
Our study was modelled on the dynamic selective-looking paradigm used by Most et al., 2001, Most et al., 2005 but we set out to create experimental conditions in which observers might develop an attentional set different from the set corresponding to the task instructions. One prerequisite was that there should be two ways to discriminate between task-relevant and task-irrelevant items, such as discriminating on the basis of shape and on the basis of colour. We used associations between shape properties and colour properties, so that task-relevant and task-irrelevant items could be discriminated in those two ways. An important consideration was whether it was more likely that observers would develop an attentional set for colour when task instructions had specified shape or that they would develop an attentional set for shape when task instructions had specified colour. There is evidence that attention is more readily guided by colour than by shape; for example, colour, more than shape, guides search for a colour-shape conjunction target (Bichot, Rossi, & Desimone, 2005). Theeuwes (1992; see also 1991) demonstrated that search for an item with a unique shape was slowed by an item with a unique colour but search for an item with a unique colour was not affected by an item with a unique shape. However, Theeuwes found that, when the colours were made so similar (yellowish red versus yellowish green) that colour discrimination was harder than shape discrimination, the pattern of results was reversed. Therefore, our hypothesis was that observers in an inattentional blindness experiment might develop an attentional set for a property that was not specified in the task instructions if it allowed easier discrimination between task-relevant and task-irrelevant items. To provide a clear case for the first investigation of this hypothesis, task instructions always specified a shape property and we ensured that the colour discrimination was substantially easier than the shape discrimination.
We used squares and diamonds as primary-task items and shape was the property explicitly specified in the task instructions. The square–diamond shape discrimination was difficult, and so we needed to confirm that it was possible for observers to use shape information to carry out the primary task of counting bounces by squares. This was not guaranteed by the fact that observers in multiple object tracking studies are able to track four of eight (or even five of ten) identical objects (Pylyshyn, 2004, Pylyshyn and Storm, 1988; see Scholl, 2009, for a review), because those observers were not required to maintain a count of bounces by the tracked objects. We conducted a pilot study with black squares and black diamonds, which confirmed that observers were able to use the square–diamond shape discrimination to carry out the primary task.
Importantly, shape and colour varied together: all four squares were black and all four diamonds were white. Thus, task-relevant items and task-irrelevant items could also be discriminated by colour. Observers who were instructed in the primary task to count bounces by squares were counting bounces by the four black squares, and observers instructed to count bounces by diamonds were counting bounces by the four white diamonds. Thus, when square was the instructed property, black was the associated property, and when diamond was the instructed property, white was the associated property. The black–white colour discrimination was substantially easier than the square–diamond shape discrimination and we predicted that observers would develop an attentional set for colour.
In three experiments, we explored the relationship between task instructions and attentional set as evidenced by the detection of (unexpected) critical stimuli. Experiment 1 (baseline) established that observers had the attentional resources required to detect a critical stimulus (an additional black square) when they were engaged in the demanding task of counting the number of times four squares bounced off the edges of the computer display. Experiment 2 was pivotal and provided evidence that observers developed an attentional set for a property that was not specified in the task instructions. Experiment 3 extended the findings of Experiment 2 and excluded an alternative explanation.
Section snippets
Observers
Seventy-five (36 male and 39 female) neurologically healthy university students with normal or corrected-to-normal vision participated (18–22 years, M = 19.8 years). Observers were randomly allocated to one of five experimental conditions, with 15 observers in each condition. The protocol was approved by the University of Oxford Research Ethics Committee, and was in accordance with the ethical standards laid down in the 2008 Declaration of Helsinki. Observers were fully debriefed on completion of
Experiment 1
Experiment 1 (Match–Match Condition) was the baseline experiment, in which 15 observers were instructed to count bounces by squares and the critical stimulus was an additional black square. The critical stimulus was reported by 12 observers (80%) in Trial 4 (critical trial). A Mann–Whitney U Test confirmed that primary-task accuracy (i.e., accuracy in counting bounces by squares) in Trial 4 was not significantly different for observers who did report the critical stimulus (N = 12, Md (median) =
Discussion
Starting from the assumption that “what you see is what you set” (Most et al., 2005, p. 217), we have shown that it is important to distinguish between the observer’s attentional set at the time the critical stimulus is presented and the attentional set corresponding to the task instructions. Observers in an inattentional blindness experiment might develop an attentional set for a property that was not specified in the task instructions. We used a dynamic selective-looking paradigm, with
Conclusion
In most studies of inattentional blindness, it is reasonable to assume that the observers adopt an attentional set for the property specified in the task instructions and maintain that set until the critical stimulus is presented. But observers might develop an attentional set for a different property if it allowed easier discrimination between task-relevant and task-irrelevant items. We found that observers who were instructed to count bounces by squares (that were black) failed to notice an
References (24)
- et al.
The role of perceptual load in inattentional blindness
Cognition
(2007) - et al.
The effects of eye movements, spatial attention and stimulus features on inattentional blindness
Vision Research
(2004) - et al.
Selective looking: Attending to visually specified events
Cognitive Psychology
(1975) - et al.
Parallel and serial neural mechanisms for visual search in macaque area 4
Science
(2005) - et al.
Involuntary covert orienting is contingent on attentional control settings
Journal of Experimental Psychology: Human Perception and Performance
(1992) - et al.
The structure of attentional control: Contingent attentional capture by apparent motion, abrupt onset, and color
Journal of Experimental Psychology: Human Perception and Performance
(1994) - et al.
How meaning shapes seeing
Psychological Science
(2007) - et al.
The role of unattended distractors in sustained inattentional blindness
Psychological Research
(2008) - et al.
The effects of perceptual load on semantic processing under inattention
Psychonomic Bulletin and Review
(2009) - et al.
Attention on autopilot: Past experience and attentional set
Visual Cognition
(2006)
Inattentional blindness
Cited by (20)
Visions of light: New reconstruction techniques of photometric data and visual perception inside Etruscan painted tombs
2023, Journal of Archaeological ScienceInattentional blindness: Attentional set for efficient task success
2023, Consciousness and CognitionThe prioritisation of motivationally salient stimuli in hemi-spatial neglect may be underpinned by goal-relevance: A meta-analytic review
2022, CortexCitation Excerpt :It is not clear, however, how intentional this effect may be, and whether participants are aware that they are tuning their attention to specific features more than others. Indeed, explicit awareness of stimulus features is not required for the adoption of an attentional set, which can be formed through implicit learning of target features independent of task instructions (Aimola Davies et al., 2013; Cosman & Vecera, 2014; Sha et al., 2017). Further, goal-driven attentional capture by motivationally salient stimuli has been found to occur when instructed to respond simply to stimulus location, without explicit responses required towards the stimulus features (Lichtenstein-Vidne et al., 2012; Vromen et al., 2016).
Inattentional blindness on the full-attention trial: Are we throwing out the baby with the bathwater?
2018, Consciousness and CognitionBeware of the gorilla: Effect of goal priming on inattentional blindness
2017, Consciousness and CognitionCitation Excerpt :For instance, Folk and Remington (2008) demonstrated that priming a specific feature (e.g., color) can nonconsciously direct visual attention in a subsequent task. Davies, Waterman, White, and Davies (2013) also demonstrated that observers can nonconsciously develop an attentional set for a property not specified in the task instructions. Here, we propose that goal priming could create a nonconscious attentional setting in implicit working memory, allowing the use of remaining attentional resources.
Brief mindfulness induction reduces inattentional blindness
2015, Consciousness and CognitionCitation Excerpt :One common cause of IB is the top-down tunings in working memory of the information processing system. The influence of these tunings is evidenced by findings that distractors that share features with targets are more likely to be detected than those which do not (Aimola Davies, Waterman, White, & Davies, 2013; Most, 2013; Most & Astur, 2007; Most et al., 2001; Simons & Chabris, 1999). Perhaps mindfulness prevents top-down information processing biases from guiding attention.