Elsevier

Brain and Cognition

Volume 71, Issue 2, November 2009, Pages 165-172
Brain and Cognition

Electrophysiological markers of categorical perception of color in 7-month old infants

https://doi.org/10.1016/j.bandc.2009.05.002Get rights and content

Abstract

The origin of color categories has been debated by psychologists, linguists and cognitive scientists for many decades. Here, we present the first electrophysiological evidence for categorical responding to color before color terms are acquired. Event-related potentials were recorded on a visual oddball task in 7-month old infants. Infants were shown frequent presentations of one color (standard) interspersed with infrequent presentations of a color that was either from the same category (within-category deviant) or from a different category (between-category deviant) to the standard. Differences in the event-related potentials elicited by the stimuli were found that were related to the categorical relationship of the standard and the deviant stimuli. The data are discussed in relation to the processes that underlie categorical responding in infancy, as well as the debate about the origin of color categories in language and cognition.

Introduction

The origin and nature of color categories in language and cognition has been the concern of researchers from a range of disciplines such as psychology, anthropology, cognitive science, linguistics and philosophy, for many decades (see Bornstein, 2006, Bornstein, 2008). One major issue has been whether the way in which the spectrum of color is carved up into categories is arbitrary, or whether there are universal constraints on how these categories form (e.g., Berlin and Kay, 1969, Kay, 2005). Variation in how languages categorize color may suggest that this process is arbitrary. For example, there is ample evidence that the number of basic color terms and the location of color category boundaries varies across languages, and that color lexicons also evolve over time (e.g., MacLaury, Paramei, & Dedrick, 2008). However, systematic investigation of over one-hundred of the world’s unwritten languages suggests there are universal constraints that underpin the color naming systems of these languages, with statistical tendencies for color categories to form at certain points in the color space (Kay and Regier, 2003, Regier et al., 2005, Regier et al., 2007).

In support of the argument that there may be perceptual constraints on how language categorizes the spectrum of color, there is converging behavioral evidence for categorical responding to color in pre-linguistic infants (Bornstein et al., 1976, Catherwood et al., 1987, Catherwood et al., 1990, Franklin and Davies, 2004, Franklin et al., 2005, Franklin et al., 2008). Four-month old infants, when habituated to a colored stimulus, dis-habituated to a novel color from a different (English) linguistic color category to the habituated stimulus, yet did not dis-habituate if the novel and habituated stimuli came from the same linguistic category (Bornstein et al., 1976).1 This effect was found for same and different-category stimulus pairs equated in wavelength-differences, for red, green, yellow and blue categories. The findings have since been replicated and extended using an alternative metric (Munsell)2 to equate same- and different-category stimulus pairs, and a novelty preference task (Franklin & Davies, 2004). These infant studies demonstrate a form of categorical responding called Categorical Perception (CP). Color CP is the faster or more accurate discrimination of two colors belonging to different categories (between-category) than two colors from the same category (within-category), even when within- and between-category chromatic separation sizes are equated (Harnad, 1987).3 Further evidence for color CP in infants comes from a target detection task, where 4–6-month old infants were faster at detecting a colored target on a different-category than same-category background, even though hue separations were the same across conditions (Franklin et al., 2005, Franklin et al., 2008).

Infants’ categorical responding to color is not surprising as categorization is a pervasive aspect of infant cognition (e.g., Mareschal & Quinn, 2001). However, that infants respond categorically to colors (such as blue and green) is surprising in the context that many of the world’s languages do not have words for these categories. The major challenge is to establish how infants form these color categories, what perceptual and cognitive processes underlie this categorical responding, and to establish how these pre-linguistic perceptual categories may (or may not) provide constraints on linguistic color categorization later on in development.

The current investigation used an ERP technique to further investigate color CP in infancy. An ERP is the voltage change, measured from electrodes on the scalp, that occurs during a defined period of time (epoch) within which a stimulus is shown or an event occurs. Using this technique we aimed to identify the electrophysiological markers of categorical responding to color in infants. There are several reasons for taking this approach. First, if color category effects in infant ERPs are found, this would provide compelling further evidence in support of the argument that color category effects exist in the absence of language. Second, as different ERP components indicate different processes (e.g., attentional allocation, novelty detection and memory updating), the technique provides information on the mechanisms that underlie the effect, as well as providing information on the time course of the effect. Third, testing for color category effects in infant ERPs extends previous research on the neural markers of CP and categorization in infancy. The ERP approach has previously been used to investigate category effects in infancy, although there have been few studies: one for CP of phonemes (Dehaene-Lambertz & Baillett, 1998) and two for category learning (Grossman et al., in press, Quinn et al., 2006). The current study allowed an assessment of whether the neural markers for category effects generalize across different perceptual domains (e.g., color/phonemes) and different types of categorical responding (e.g., categorical perception/perceptual categorization). Whereas categorical perception requires that same- and different-category differences are equated, perceptual categorization such as that in Quinn et al.’s study does not. However, despite these differences, there may well be similarities in the neural markers and underlying mechanisms of these two types of categorical responding.

In the current investigation, infant ERPs were recorded during a visual oddball task. The task involves frequent presentation of one stimulus (the standard), while another stimulus or other stimuli are presented less frequently (the ‘oddball’ or deviant; e.g., de Haan & Nelson, 1997). An oddball effect is found when there are differences between the deviant and frequent stimulus in the amplitude or latency of ERP components. The design of the current study follows that of an adult ERP study of color CP (Holmes, Franklin, Clifford, & Davies, 2009), where ERPs were recorded to deviant stimuli that were either from the same category (within-category) or from a different category (between-category) to the standard, with within- and between-category hue differences equated. In the adult study, ERP components peaked earlier and were of greater amplitude for between-category deviants compared to within-category deviants, even for very early perceptual components (P1 and N1; Holmes et al., 2009).

In infants, ERP components that are typically elicited on a visual oddball task are the Nc, the negative slow wave (NSW) and the positive slow wave (PSW; e.g., de Haan & Nelson, 1997). The Nc is a typically central negativity that peaks around 400 ms, and is commonly interpreted as a marker of attentional allocation (e.g., Richards, 2003), with pre-frontal and anterior cingulate origins (Reynolds & Richards, 2005). The positive and negative slow waves are thought to reflect more diffuse activation of neural areas, with the NSW seen as reflecting the detection of novel events against a background of familiar events, and the PSW reflecting the extent of stimulus encoding and updating in working memory (e.g., Nelson & Monk, 2001). Category effects were found in these visual ERP components for category learning of cats and dogs in 6-month old infants (Quinn et al., 2006). The current study tests whether category effects in these ERP components generalize to categorical responding for another domain and type of categorical responding.

In addition to the visual oddball task, a control task was included where ERPs were recorded to alternations of deviant stimuli in the visual oddball task. This ensured that any differences in the ERP components for the within- and between-category deviants could be attributed to the categorical status of the deviant in relation to the standard, rather than other stimulus characteristics, such as a greater salience of one hue over the other. The standard stimulus was never shown, thereby removing the categorical context of the stimuli. To aid the interpretation of any effects found in the ERP components, a behavioral post-test followed the visual oddball and control task, where looking time was recorded to the two stimuli that were deviant in the visual oddball task and that were alternately presented during the control task.

Section snippets

Participants

Fifty-nine 7-month old infants took part, with 20 infants not included in the final sample due to general inattentiveness (n = 5) or a high number of movement artifacts in the ERP data (n = 15). Infants were allocated to either visual oddball or control tasks. The visual oddball task had a final sample of 20 infants (9 female and 11 male) with a mean age of 236 days (SD = 17) and the control task had a final sample of 19 infants (10 female and 9 male) with a mean age of 242 days (SD = 16). All birth

Visual oddball task

Time windows for the Nc (250–650 ms) and the NSW/PSW (1150–1700 ms) were determined on the basis of prior research and inspection of each individual’s waveforms. Fig. 3 shows the waveforms for each type of stimulus, with the Nc and NSW/PSW time frames illustrated for C3 and C4. The waveforms are given for all electrodes except Oz which was excluded from the analysis due to high levels of noise.

Discussion

On a visual oddball task, infant ERPs were recorded to infrequent deviant stimuli that were either from the same or different color category to a frequently presented standard stimulus. Analysis of stimulus differences within the Nc and NSW and PSW time ranges revealed differences that were consistent with color CP and categorical responding to color. First, the between-category deviant elicited a greater Nc than the within-category deviant, with an oddball effect (greater Nc for deviant than

Acknowledgments

Research was supported by a Joint Innovation grant to Anna Franklin, Amanda Holmes and Ian Davies and a departmental Ph.D. bursary to Alexandra Clifford. We thank Laura Bevis, Justine Cornforth, Gilda Drivonikou and Jacqueline Hood for infant testing assistance, and Alex Hogan for advice on infant EEG. We are grateful to the infants and their parents.

References (42)

  • M.H. Bornstein

    Hue categorization and color naming: Cognition to language to culture

  • M. Bornstein et al.

    Color vision and hue categorisation in young human infants

    Journal of Experimental Psychology: Human Perception and Performance

    (1976)
  • M. Bornstein et al.

    Discrimination and matching within and between hues measured by reaction times: Some implications for categorical perception and levels of information processing

    Psychological Research

    (1984)
  • D. Catherwood et al.

    The nature of infant memory for hue

    British Journal of Developmental Psychology

    (1987)
  • D. Catherwood et al.

    The course of infant memory for hue

    Australian Journal of Psychology

    (1990)
  • T. DeBoer et al.
  • M. de Haan et al.

    Recognition of the mother’s face by six-month-old infants: A neurobehavioral study

    Child Development

    (1997)
  • G. Dehaene-Lambertz et al.

    A phonological representation in the infant brain

    NeuroReport

    (1998)
  • G.V. Drivonikou et al.

    Further evidence for lateralization of Whorfian effects to the right visual field

    Proceedings of the National Academy of Sciences

    (2007)
  • A. Franklin et al.

    New evidence for infant colour categories

    British Journal of Developmental Psychology

    (2004)
  • A. Franklin et al.

    Categorical Perception of color is lateralized to the right hemisphere in infants, but to the left hemisphere in adults

    Proceedings of the National Academy of Sciences

    (2008)
  • Cited by (39)

    • Color categorization in infants

      2019, Current Opinion in Behavioral Sciences
      Citation Excerpt :

      Similarly, Franklin and Davies [12•] found novelty preferences in infants (4–6 months) dividing green from blue, blue from purple and pink from red. Event-related potentials (ERP) have also been used to demonstrate infants’ categorical response to color at seven months, revealing differences in ERP components related to attentional allocation [13]. Importantly, in these and other infant studies in this review, the infants taking part are pre-linguistic — they do not yet produce language nor show comprehension of color terms [14•].

    • Emergence and separation of color categories: an NIRS study in prelingual infants and a k-means analysis on Japanese color-naming data

      2019, Current Opinion in Behavioral Sciences
      Citation Excerpt :

      Given that the use of words is not necessary, considering the farmer’s example and the presence of color categories in non-human primates (the latter will be explained later in details) [28–30], color categories could already exist in the brains of human infants. It follows that, although several behavioral studies reported the presence of color categories in pre-lingual infants [6–11], not many attempts were made to measure the cortical activity related to color categories in infants. A noninvasive brain-imaging technique, referred to as near-infrared spectroscopy (NIRS), was employed in the current investigation [4••], as it enables the estimation of the brain activity through the measurement of neurons’ oxygen consumption.

    • The biological bases of colour categorisation: From goldfish to the human brain

      2019, Cortex
      Citation Excerpt :

      The passive oddball task led the observers’ attention away from colours, toward an unrelated cover task. Some of the studies using this kind of task found category effects on early components (Clifford et al., 2010; Forder et al., 2017b; Thierry et al., 2009), while others did not, at least not in infants (Clifford et al., 2009). In contrast, the active oddball task and the attentional blink task (Maier & Rahman, 2018) required observers to focus on the colours.

    • Rethinking foundations of language from a multidisciplinary perspective

      2018, Physics of Life Reviews
      Citation Excerpt :

      However, the relativist hypothesis states that linguistic categorization results mainly from linguistic and socio-cultural factors [202–205] and forces like grouping perceptually-close stimuli has a loose effect on linguistic categorization [206,207]. Many approaches have been adopted to address this issue, such as psychological experimentation of color acquisition [208–212] and perception [213–216], neuroimaging studies of categorical perception of color stimuli [217–221], and agent-based simulations of color category development and spread [222–227]. These studies report contrasting results in support of either of the two hypotheses.

    View all citing articles on Scopus
    View full text