Categorical influences on chromatic search asymmetries

Abstract

The experiments reported here investigate asymmetries in performance of speeded target detection and visual search tasks using colourful stimuli. Experiment 1 found evidence of asymmetries in the performance of within-category trials in both tasks. Two further experiments consider possible explanations. The results of Experiment 2 suggest these asymmetries are not linked to changes in the frequency with which individual stimuli appear in trials. A third experiment shows that asymmetries in performance of the two tasks are accompanied by changes in stimulus categorization, thus providing evidence that asymmetries in performance of search tasks may represent a language driven category effect. In the Discussion section, theories describing how category information might give rise to the asymmetries found are considered.

Keywords:

• Categorization
• Category Effects
• Colour Perception
• Visual Search

Acknowledgments

I am grateful for the help provided by Pınar Boyacı and Mahperi Uluyol in data collection, and also for the constructive comments made by three reviewers.

Notes

¹The term categorical perception is often used to describe this phenomenon. However, it is not clear whether such effects involve perceptual processes. Consequently, the term category effect is used throughout the text.

²There is also some evidence that Turkish speakers divide the region of colour space categorized as “blue” by English speakers in two separate basic colour terms, with the term “lacivert” corresponding to “navy” blue and “mavi” for other blue hues (see Özgen & Davies, 1998). The stimuli included in this experiment all fall within regions of colour space named “mavi” or “yesil” (green) by Turkish speakers.

³Additional analyses carried out using data from the visual search task investigated possible lateralized asymmetries in task performance. Using similar tasks, Gilbert et al. (2006) and Roberson et al. (2008) reported that the across-category advantage is either absent for stimuli presented to the left visual field or more pronounced for stimuli presented to the right visual field. In one analysis, following the procedure used in previous studies, data from within-category trials was collapsed prior to comparison with across-category trials. Analysis was with a 2 (visual field: Left, right) × 2 (category: Within, across) repeated measures ANOVA. A significant main effect of category was apparent, F(1, 16) = 90.6, MSE = 4886, p < .001, with responses faster on across than within-category trials. The main effect of visual field approached but did not reach significance, F(1, 16) = 3.83, MSE = 4560, p = .068. Critically, the interaction failed to reach significance, F(1, 16) = 0.07, MSE = 3523, p = .788. A second analysis compared data from the two within-category pairs using a 2 (visual field: Left, right) × 2 (pair: Within “green”, within “blue”) repeated measures ANOVA. As in the first analysis, there was no main effect of visual field, F(1, 16) = 1.54, MSE = 8751, p = .233, and no interaction, F(1, 16) = 0.08, MSE = 2440, p = .779. The results of these analyses appear to support the results of other research (Brown et al., 2011; Witzel & Gegenfurtner, 2011) that failed to find evidence of lateralized asymmetries in colour categorical effects.

⁴This author is aware of no studies that have directly compared differences in responding in visual search and target detection tasks of the kind used to assess colour category effects. Although several studies have used one or other task to assess colour categorical responding, direct comparison of results is complicated by the fact that few studies have used identical stimuli pairs in both task types. This makes it difficult to isolate task driven performance differences from performance differences that arise due to differences caused by use of different stimulus pairs. One exception is a study carried out by Witzel and Gegenfurtner (2011) that used both visual search and target detection tasks to assess colour categorical responding. Two of the stimulus sets they used were employed in both visual search and target detection tasks. One stimulus set consisted of three stimulus pairs spanning the green–blue region of colour space (this set was used in Implementation 1c of their visual search task and in Implementation 2a of their target detection task). For this set, overall response times were slower on the visual search than on the target detection task. Also the “category” effect, which they calculated by comparing trials involving across to within-category stimulus pairs, appears to have been larger in the visual search than in the target detection task: That is, response times were approximately 20% faster on across than on within-category trials on the visual search task. The equivalent across-category advantage for the target detection task using the same stimulus set was 4% (these figures are derived from Tables S5–S7 in Witzel and Gegenfurtner's supplementary material). A second stimulus set they used in both tasks was drawn from the blue–purple region. Again, response times appear to have been higher in the visual search than target detection task, and the category effect also appears more pronounced in the visual search task (see same tables). These results appear to resemble to the results of Experiment 1 reported in the main text: In Experiment 1 responses were slower overall in the visual search than in the target detection task, whereas the category effects were relatively larger in the visual search task. It should be noted that the “category” effects Witzel and Gegenfurtner report do not involve within-pair asymmetries in responding and that they conducted no statistical analysis comparing the size of the category effects in different tasks. It should further be noted that Witzel and Gegenfurtner ascribe their “category” effects to differences in perceptual discriminability between stimulus pairs and thus claim they do not represent genuine category effects.