Elsevier

Cognition

Volume 124, Issue 2, August 2012, Pages 107-116
Cognition

Intentional subitizing: Exploring the role of automaticity in enumeration

https://doi.org/10.1016/j.cognition.2012.05.010Get rights and content

Abstract

Subitizing is traditionally described as the rapid, preattentive and automatic enumeration of up to four items. Counting, by contrast, describes the enumeration of larger sets of items and requires slower serial shifts of attention. Although recent research has called into question the preattentive nature of subitizing, whether or not numerosities in the subitizing range can be automatically accessed is yet to be empirically tested. In the current study, participants searched for two pre-defined digits in a circular visual-search array. Distractor dots of various set sizes were placed at the centre of the array. Despite the relevance of the distractor numerosities to the target detection task, the distractors did not influence target detection, thereby suggesting that their numerosities were not automatically accessed in Experiment 1. In Experiment 2, participants were explicitly instructed to enumerate the distractor dots. Here, congruent and incongruent distractor numerosities influenced the target detection task, thereby revealing that the distractor dots were capable of generating interference. Experiment 3 ensured that dots were attended by asking participants to detect the luminance of dots. Data confirmed that subitizing was not automatic. The present study also supported the alleged discontinuity between the subitizing and counting ranges because an examination of reaction time gradients in Experiment 2 found the counting gradient to be significantly steeper than the subitizing gradient. In sum, the results suggest that subitizing is a distinct but non-automatic style of enumeration.

Highlights

► We empirically investigate the role of automaticity in enumeration. ► We show that, contrary to current theories, subitizing is not an automatic process. ► Access to numerical information requires an explicit intention to enumerate.

Introduction

The enumeration of visual stimuli appears to be underpinned by two distinct psychological processes. Although lay understandings of numerical cognition suggest that children are explicitly taught to “count” three images on a page, psychological research dictates that counting is a process reserved for the enumeration of more than four objects. By contrast, enumeration of four or fewer objects is referred to as subitizing (Chi and Klahr, 1975, Kaufman et al., 1949, Mandler and Shebo, 1982). The distinction between subitizing and counting has been motivated by the finding that people almost instantly access the numerosity of displays containing four or fewer items, implying that the numerosity is automatically available without the need for a serial individuation of each item (Trick and Pylyshyn, 1993, Trick and Pylyshyn, 1994). Reaction times (RTs) are typically flat or reveal a shallow gradient as the number of display items increase across the subitizing range (Akin and Chase, 1978, Chi and Klahr, 1975, Jensen et al., 1950, Mandler and Shebo, 1982, Piazza et al., 2002, Sathian et al., 1999, Trick and Pylyshyn, 1994). When displays contain more than four items, enumeration is not automatic and participants must index each item individually in a step-by-step manner. RT gradients are consequently much steeper in the counting range, creating a discontinuity between the subitizing and counting ranges (Jensen et al., 1950, Mandler and Shebo, 1982, Saltzman and Garner, 1948). Interestingly, subitizing is not limited to the enumeration of visual objects by adults, but exists in auditory and tactile domains (Camos and Tillman, 2008, Plaisier et al., 2009, Riggs et al., 2006), and occurs in pre-verbal infants and animals (Davis and Perusse, 1988, Spelke, 2000, Starkey and Cooper, 1980). Some researchers argue that counting and subitizing are phenomenologically distinct processes because subitizing reflects a preattentive, automatic process that proceeds via parallel processing whereas counting is a more attentionally demanding, conscious mechanism that implicates serial processing (Trick and Pylyshyn, 1993, Trick and Pylyshyn, 1994). The current study focused on investigating the role of automaticity in subitizing.

One of the most popular explanations of enumeration – the fingers of instantiation (FINST) hypothesis – presumes that subitizing occurs, at least in part, via automatic, preattentive mechanisms (Trick and Pylyshyn, 1993, Trick and Pylyshyn, 1994). FINST is rooted in the notion that the visual system is capable of indexing up to four objects before capacity limitations ensue. According to this hypothesis, four FINSTs can be ‘attached’ to four objects, individuating the objects across space. This process is both goal driven and data driven: it is automatically driven by stimuli in the visual field, but is also influenced by the individual’s intentions and goals. In other words, aspects of subitizing (for example, the parallel attentive and token indexing stages) are automatic, yet enumeration itself is a deliberate process – people do not automatically enumerate everything that they perceive (Trick & Pylyshyn, 1994). The FINST hypothesis suggests that enumeration of small numbers is rapid because objects are tagged in parallel and the participant simply has to report the number of FINSTs that are bound at that time. Enumeration of more than four items necessitates slower shifts of spatial attention because FINSTs must be serially connected, disconnected, and re-connected until all objects have been tagged and counted. The current study was interested in exploring the nature of automatic processing within subitizing.

Although the role of automaticity in subtizing has not been previously explored, recent research has challenged the notion of subitizing as a preattentive process (Burr et al., 2010, Egeth et al., 2008, Olivers and Watson, 2008, Poises et al., 2008, Railo et al., 2008, Vetter et al., 2008, Xu and Liu, 2008). These studies have revealed that attentional demands increase with increasing numbers of to-be-enumerated stimuli, even in the subitizing range. For example, a few experiments have incorporated a subitizing task into the attention blink paradigm in order to explore the relationship between subitizing and attention (Egeth et al., 2008, Olivers and Watson, 2008, Xu and Liu, 2008). The attentional blink describes a deficit in detecting the second of two targets when the second target appears within 500 ms of the first target (Raymond, Shapiro, & Arnell, 1992). Accuracy is poor for this second target because attentional resources are unavailable (Chun and Potter, 1995, Shapiro et al., 1994). If subitizing proceeds preattentively then enumeration should be accurate even when a subitizing task is inserted into the attentional blink paradigm during periods of reduced attentional capacity. However, the empirical evidence counters this logical suggestion. Specifically, the enumeration of small numerosities was found to be significantly delayed or suffered reduced accuracy inside the attentional blink period, hence implicating attention as a critical requirement for subitizing (Egeth et al., 2008, Olivers and Watson, 2008, Xu and Liu, 2008). Given the challenges to the preattentive nature of subitizing, and considering the connection between attention and automaticity, we suggest that the assumption of automaticity within subitizing may also be problematic.

Automaticity evades a precise theoretical definition. Numerous descriptions of automatic processing have been suggested, including processing that occurs without conscious control, intention or awareness (Logan, 1988, Schneider and Shiffrin, 1977, Shiffrin and Schneider, 1977). In addition, formal definitions of automaticity have been provided by a number of authors (Gibson, 1969, LaBerge, 1973, Logan, 1992, Treisman et al., 1992). Automaticity can be contrasted with processing that is conscious, effortful, intended and controlled. Automatic processing is frequently equated with preattentive processing, and empirical evidence for interactions between the two mechanisms suggest that they are related mechanisms at minimum, and ‘different sides of the same coin’ at maximum (Logan, Taylor, & Etherton, 1999). Further, Bargh and Chartrand (1999) suggest that preattentive processing is one of the ‘major strains’ of automaticity. However, while preattentive and automatic processes are related (primarily because they can both occur in the absence of attention), the processes are not identical. For example, the way in which automatic and preattentive processes become independent of attention differs (Treisman et al., 1992). The recent challenge to subitizing as a preattentive process, and the link between preattentive processing and automaticity, suggests that it is prudent to explicitly examine the status of subitizing as an automatic process. Given the similarities between preattention and automaticity, it may be the case that subitizing is not automatic because it is not preattentive. Conversely, the dissociability between these processes means that aspects of subitizing may proceed automatically, but not preattentively.

It is important to note that the current study does not seek to explore whether enumeration proceeds automatically – irrespective of context. It would be unwise to assume that people constantly and persistently enumerate every visual object that they encounter. Similarly, the FINST hypothesis does not suggest that subitizing (or enumeration more generally) is absolutely automatic. While FINST argues that up to four items can be indexed automatically, the mapping of numerosity to number names is not presumed to be automatic but requires a controlled, conscious intention. The current study therefore asked whether the numerical value of a group of dots would be automatically accessed when participants performed a number detection task but were not asked to name the number of dots. In other words, we asked whether numerical value would be accessed when the contextual conditions were highly conducive to such numerical access.

This study required participants to search a circular visual search array and detect whether the Arabic digit “3” or “7” was present. In this manner, “3” and “7” served as targets. The task was designed so as to include one target from the subitizing range and one from the counting range. Distractor stimuli were presented in the centre of the search array and consisted of either two, three, six or seven dots. Distractors were located at fixation in order to ensure that they were positioned within participants’ visual fields. If subitizing proceeds automatically, presentation of the three-dot distractor should influence responding by speeding RTs on congruent target-3 trials and slowing RTs on target-7 trials (Henik and Tzelgov, 1982, Rubinsten et al., 2002, Soltesz et al., 2010, Soltesz et al., 2011, Szűcs et al., 2007). According to our assumptions, enumeration of numerosities in the counting range should not proceed automatically. Therefore, we did not expect to see interference from seven-dot distractors. The neutral two-dot and six-dot distractors were employed in order to prevent participants from anticipating the three or seven dot distractors (the two-dot and six-dot distractors were neutral because ‘2’ and ‘6’ were not target stimuli). Two and six dots were also chosen because they are each very close to the target digits, and fall within the subitizing and counting ranges respectively. Inclusion of the neutral distractors saw that we were able to calculate a gradient for the subitizing range distractors (the gradient between two-dot and three-dot distractor trials) and for the counting range distractors (the gradient between six-dot and seven-dot distractor trials).

We also manipulated attentional load in this study (see Vetter et al., 2008). On high load trials, the target digit was presented amongst numerous different visual-search stimuli whereas the target appeared amongst a single repeated stimulus on low load trials. The perceptual load theory predicts that interference should only occur on low load trials, where attentional resources are available, and not on high load trials where attentional resources are taxed (Lavie, 1995, Lavie and Tsal, 1994). However, if subitizing is preattentive then interference should occur across both high and low loads. Because recent research suggests that subitizing is not preattentive but requires attentional resources, we predicted that interference would be restricted to low load trials, where attentional resources are available to process the distractor dots. In sum, the current experimental design allowed us to investigate three main issues: whether subitizing proceeds automatically, whether subitizing is preattentive, and whether the subitizing and counting gradients are statistically distinct.

Section snippets

Experiment 1

Experiment 1 was primarily designed to investigate whether the numerosity of dots in the subitizing range can be automatically accessed, and cause interference on a numerical target detection task. If activation of the numerosity is automatic, it should interfere with target detection due to congruent and incongruent relationships between the target and distractor numerosities. As discussed in Section 1, this experiment did not ask about the automaticity of enumeration in general (under any

Experiment 2

Experiment 2 reacted to the finding that the distractor dots did not interfere with the target detection task in Experiment 1. We therefore questioned whether distractor dots are ever able to influence RTs in this paradigm. To that end, this experiment was identical to Experiment 1 with one exception; participants were requested to verbally enumerate the distractor dots after they responded to the target detection task. If the distractor dots successfully generate interference in this task, the

Experiment 3

Experiments 1 and 2 collectively suggest that participants accessed the distractor dots’ numerical values only when there was an explicit intention for enumeration. However, it is important to rule out alternative interpretations of these findings. For example, although we argue that attention was directed towards the distractor dots in Experiment 1 (given their location at fixation – see Section 1), we cannot guarantee that the distractors were attended in that experiment. It is therefore

General discussion

The current study examined the role of automaticity in subitizing by presenting distractor dots within a numerical target detection task. The results suggested that accessing numerical information in the subitizing range is not automatic because the numerosity of the distractor dots was only accessed when participants were explicitly asked to enumerate those stimuli. Our study also contributed to the ongoing controversy concerning whether or not subitizing and counting are distinct phenomena.

Acknowledgements

This research was funded by a Grant to DS from the Medical Research Council (G0900643), and a Gates Cambridge studentship to HP. The authors would like to thank Alison Nobes for her assistance in collecting data for these experiments.

References (57)

  • J. Balakrishnan et al.

    Is subitizing a unique numerical ability?

    Perception and Psychophysics

    (1991)
  • J. Balakrishnan et al.

    Subitizing: Magical numbers of mere superstition?

    Psychological Research

    (1992)
  • J. Bargh et al.

    The unbearable automaticity of being

    American Psychologist

    (1999)
  • D.M. Beck et al.

    Look here but ignore what you see: Effects of distractors at fixation

    Journal of Experimental Psychology: Human Perception and Performance

    (2005)
  • D. Burr et al.

    Subitizing but not estimation of numerosity requires attentional resources

    Journal of Vision

    (2010)
  • M.M. Chun et al.

    A two-stage model for multiple target detection in rapid serial visual presentation

    Journal of Experimental Psychology: Human Perception and Performance

    (1995)
  • H. Davis et al.

    Numerical competence in animals: Definitional issues, current evidence, and a new research agenda

    Behavioural and Brain Scieneces

    (1988)
  • S. Dehaene et al.

    Dissociable mechanisms of subitizing and counting: Neuropsychologial evidence from simultagnosic patients

    Journal of Experimental Psychology: Human Perception and Performance

    (1994)
  • N. Demeyere et al.

    Neuropsychological evidence for a dissociation in counting and subitizing

    Neurocase

    (2010)
  • H. Egeth et al.

    The role of attention in subitizing: Is the magical number 1?

    Visual Cognition

    (2008)
  • C.L. Folk et al.

    Involuntary covert orienting is contingent on attentional control settings

    Journal of Experimental Psychology: Human Perception and Performance

    (1992)
  • S. Forster et al.

    Failures to ignore entirely irrelevant distractors: The role of load

    Journal of Experimental Psychology: Applied

    (2008)
  • E.J. Gibson

    Principles of perceptual learning and development

    (1969)
  • H. Haladjian et al.

    Enumerating by poitning to locations: a new method for measuring the numerosity of visual object representations

    Attention Perception and Psychophysics

    (2011)
  • A. Henik et al.

    Is three greater than five: The relation between physical and semantic size in comparison tasks

    Memory and Cognition

    (1982)
  • E. Jensen et al.

    The subitizing and coutning of visually presented fields of dots

    Journal of Psychology

    (1950)
  • E. Kaufman et al.

    The discrimination of visual number

    American Journal of Psychology

    (1949)
  • D. LaBerge

    Attention and the measurement of perceptual learning

    Memory and Cognition

    (1973)
  • Cited by (9)

    • Neuro-cognitive mechanisms of global Gestalt perception in visual quantification

      2018, NeuroImage
      Citation Excerpt :

      Moreover, fMRI studies revealed stronger activity in TPJ areas for tasks with high attentional demands like visual search (Himmelbach et al., 2006) or tasks requiring attentional shifts in more demanding visual conditions (Doricchi et al., 2010; Thiel et al., 2004). However, these results indicate that differences in attentional processes do rather not explain the detected differences between subitizing and estimation in Gestalt ROIs because subitizing is considered a pre-attentive, top-down process requiring less attention (Piazza et al., 2003; Sathian et al., 1999; Trick and Pylyshyn, 1994 but see Pincham and Szucs, 2012; Railo et al., 2008). In contrast, estimation, which requires more attentional resources due to higher task complexity, elicited lower neuronal responses in our Gestalt ROIs.

    • Effect of emotions on temporal attention

      2017, Progress in Brain Research
      Citation Excerpt :

      High arousal emotional pictures as distractors in EIB studies (Kennedy et al., 2014) were associated with a trade-off in the amplitudes of N2–P3b components. The N2 component is known to reflect attentional orienting toward the target, while P3 is known to reflect consolidation in working memory (Kennedy et al., 2014; Pincham and Szucs, 2012; Reiss et al., 2008; Sergent et al., 2005; Shapiro et al., 2006). At shorter lags, enhanced P3b amplitude is observed for emotional distractors and consequently reduced P3b during the blink period when the target is missed, whereas in trials where target is reported P3b is enhanced in magnitude in comparison to the distractor.

    • The goal-dependence of level-1 and level-2 visual perspective calculation.

      2020, Journal of Experimental Psychology: Learning Memory and Cognition
    View all citing articles on Scopus
    View full text