Elsevier

Neuropsychologia

Volume 48, Issue 9, July 2010, Pages 2769-2772
Neuropsychologia

Brief communication
Varying the scope of attention alters the encoding of categorical and coordinate spatial relations

https://doi.org/10.1016/j.neuropsychologia.2010.04.027Get rights and content

Abstract

Two types of representations can be used to specify spatial relations: Coordinate spatial relations representations specify the precise distance between two objects, whereas categorical spatial relations representations assign a category (such as above or below) to specify a spatial relation between two objects. Computer simulation models suggest that coordinate spatial relations representations should be easier to encode if one attends to a relatively large region of space, whereas categorical spatial relations should be easier to encode if one attends to a relatively small region of space. We tested these predictions. To vary the scope of attention, we asked participants to focus on the local or global level of Navon letters, and immediately afterwards had them decide whether a dot was within 2.54 cm of a bar (coordinate judgment) or was above or below the bar (categorical judgment). Participants were faster in the coordinate task after they had just focused on the global level of a Navon letter whereas they were faster in the categorical task after they had just focused on the local level. Although we did not test the hemispheric lateralization of these effects, these findings have direct implications for theories of why the cerebral hemispheres differ in their relative ease of encoding the two kinds of spatial relations.

Introduction

Researchers in cognitive psychology and cognitive neuroscience have often shown that what intuitively may seem to be a single ability, such as memory or perception, actually comprises multiple specialized functions. Such research has further shown that even what appear to be individual specialized functions often can be further subdivided. One example of such research has shown that spatial relations can be encoded in more than one way, by processes that encode coordinate versus categorical spatial relations representations (Kosslyn, 1987, Kosslyn, 2006). Coordinate representations preserve the precise metric distance between objects and specify their locations within a coordinate system. In contrast, categorical representations assign a category, such as “left of,” “above,” or “behind,” to characterize the spatial relation between objects or parts of an object. The two types of spatial relations serve different functions. Coordinate spatial relations representations are used to guide one's actions, such as reaching and manipulating an object or navigating efficiently in an environment. In contrast, categorical spatial relations representations are used to recognize shapes that are contorted in an unfamiliar way, by preserving the category of relations among parts (e.g., an upper arm and forearm remain “connected by a hinge” no matter how they are positioned; Laeng, Carlesimo, Caltagirone, Capasso, & Miceli, 2002).

A growing number of findings from behavioral (e.g., Hellige and Michimata, 1989, Kosslyn et al., 1989), neuroimaging (e.g., Baciu et al., 1999, Kosslyn et al., 1998, Slotnick and Moo, 2006), transcranial magnetic stimulation (TMS, e.g., Trojano, Conson, Maffei, & Grossi, 2006), and lesions studies (e.g., Laeng, 1994, Palermo et al., 2008) document that the brain computes categorical spatial relations more efficiently in the left cerebral hemisphere whereas it computes coordinate spatial relations more efficiently in the right cerebral hemisphere (for reviews, see Jager and Postma, 2003, Kosslyn, 1987, Kosslyn, 2006). Although researchers now generally agree that the two types of spatial relations are processed more efficiently in different hemispheres, the cause of this hemispheric specialization remains to be determined. In the present study, we generated predictions on the basis of what is known about the hemispheric specialization of spatial relations representation and processing.

One explanation for the observed lateralization of spatial relations processing hinges on differences in the sizes of regions of space attended to by the two hemispheres. Kosslyn, Chabris, Marsolek, and Koenig (1992) report a series of neural-network computer simulations in which they showed that (a) networks receiving input from units with small non-overlapping receptive fields computed categorical spatial relations representations more effectively than coordinate spatial relations representations, and that (b) networks receiving input from units with large overlapping receptive fields computed coordinate spatial relations representations more effectively than categorical spatial relations representations. Small non-overlapping receptive fields served to carve space into small bins, and the relations among these bins were easily specified; in contrast, large overlapping receptive fields can use coarse coding to register metric information about the input (e.g., O’Reilly, Kosslyn, Marsolek, & Chabris, 1990). Jacobs and Kosslyn (1994), using more sophisticated models, replicated these findings.

And in fact, researchers have presented evidence that the hemispheres do differ in the scope of space that typically is efficiently encoded. For example, lesions of the right temporo-parietal junction impair selectively the processing of an overall pattern whereas lesions of the left temporo-parietal junction impair selectively the processing of component parts of a pattern (e.g., Lamb, Robertson, & Knight, 1989). In addition, using a divided-visual-field method, Kosslyn, Anderson, Hillger, and Hamilton (1994) found that normal participants could compare two diagonal lines that were far apart more easily when the stimuli were presented briefly in the left visual field (and hence initially encoded by the right hemisphere) than when they were presented briefly in the right visual field (and hence encoded initially by the left hemisphere). This finding makes sense if the right hemisphere registers input from larger regions of space than does the left. Similarly, people generally can compare relatively low spatial frequency displays better when they are presented briefly to the left visual field, and relatively high spatial frequency displays better when they are presented briefly to the right visual field (e.g., Christman, 1997).

In the experiment reported here, we investigate whether attending to relatively large regions of space facilitates encoding coordinate spatial relations representations more than encoding categorical spatial representations, whereas attending to relatively small regions of space has the reverse effect. In order to prime participants’ attention to relatively large or small regions of space, we used Navon (1977) stimuli; these stimuli consist of large letters that are composed of many small letters (see Fig. 1a). We asked participants to make a decision based either on the large (global) letter or on the small (local) letter. After each presentation of a Navon stimulus, we presented a horizontal bar and dot (see Fig. 1b). In the categorical task, participants were asked to determine whether the dot was above or below the bar. In the coordinate task, participants decided whether the dot was located within 2.54 cm (one inch) from the bar (this task was introduced by Hellige & Michimata, 1989). If encoding categorical spatial relations representations is achieved by delineating small discrete regions in space, then participants should be more efficient in the categorical task when their attention had just been set in the Navon task to focus on small, local areas rather than large, global areas. And if encoding coordinate spatial relations representations is achieved by coarse coding, then participants should be more efficient in the coordinate task when their attention had just been set to focus on large, global areas than on small, local areas.

Section snippets

Participants

Thirty-six right-handed adults with normal or corrected-to-normal vision (20 females and 16 males with a mean age of 21 years and 2 months) from Harvard University and the local community volunteered to participate for pay or course credit. Data from one additional participant were not analyzed because he performed at chance levels, and hence we had no reason to believe that he actually performed the tasks. Participants were randomly assigned to the global or local conditions of the Navon task

Results

We analyzed separately the data from the Navon task and from the spatial relations tasks. All analyses of RTs included only data from trials on which participants responded correctly. Outliers were defined as RTs greater than 2 SDs from the mean for that participant. Outliers occurred on 1.2–1.8% of the trials in the different tasks. After removing outliers, for each participant, the average RTs for the Navon task, the categorical task and the coordinate task were computed.

As a first step, we

Discussion

As predicted, participants encoded categorical spatial relations faster if they had just focused on local letters in the Navon task, whereas they encoded coordinate spatial relations faster if they had just focused on global letters in the Navon task. The results are consistent with neural-network simulations (Jacobs and Kosslyn, 1994, Kosslyn et al., 1992) indicating that coordinate and categorical judgments operate most efficiently on outputs from units with, respectively, large or small

Acknowledgements

This research was supported by Grant R01 MH060734 from the National Institute of Mental Health. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Institute of Mental Health. We wish to thank Coralie Eggeling for her help in recruiting participants and collecting data.

Cited by (18)

  • Accelerating reading via local priming

    2020, Acta Psychologica
    Citation Excerpt :

    Hence overall, the current findings suggest a positive effect of local-directed attention on both word-level and text-level reading. The current results first support the general notion that local and global priming can have a different automatic effect on a variety of tasks (e.g., Borst & Kosslyn, 2010; Friedman et al., 2003; Macrae & Lewis, 2002; Michimata et al., 2011; VanDellen, Sanders, & Fitzsimons, 2012) and extend their influence to the reading domain as well. As mentioned in the introduction section, there might be a connection between local/global processing and reading processes.

  • On Inter- and Intrahemispheric Differences in Visuospatial Perception

    2017, Neuropsychology of Space: Spatial Functions of the Human Brain
  • Effects of the global and local attention on the processing of categorical and coordinate spatial relations

    2011, Brain and Cognition
    Citation Excerpt :

    As predicted, the coordinate task was better-performed after responding to the global level of a Navon stimulus, while the categorical task was better-performed after responding to the local level. Borst and Kosslyn’s (2010) study is an interesting precedent for the present study. However, the priming method used in that study does not guarantee that the same effect can generalize to fast or immediate change in the scope of attention.

View all citing articles on Scopus
View full text