Elsevier

Brain Research

Volume 1124, Issue 1, 8 December 2006, Pages 96-99
Brain Research

Research Report
Perceptual biases in chimeric face processing: Eye-movement patterns cannot explain it all

https://doi.org/10.1016/j.brainres.2006.09.069Get rights and content

Abstract

Experiments using chimeric faces typically report a perceptual bias towards the viewers left. Here we show that this leftward bias can be elicited even when eye movement should be impossible. Although supporting previous studies arguing that eye movements are not necessary to generate the bias, the effect we found here was significantly reduced, compared to an earlier study which allowed eye movements. We suggest that the chimeric face bias is enhanced by eye movements.

Introduction

Since Wolff (1933) first observed that the right side of the human face, that is the side that falls in the left visual field during face to face encounters, shows more correspondence to ‘the impression caused’ by the full face, the finding of a robust left perceptual bias for chimeric images has been reported with chimeric stimuli employing emotion (David, 1993, Ferber and Murray, 2005), gender (Luh et al., 1991, Butler et al., 2005), age and attractiveness (Burt and Perrett, 1997). This effect is so robust it can be demonstrated via the internet (Rueckert, 2005) and even with inverted stimuli (Butler and Harvey, 2005). Moreover, Brady et al. (2005) have recently provided evidence that such perceptual asymmetries are held in long term memory (in this case regarding the identification of very familiar chimeric faces).

It is generally assumed that these effects are due to a right hemisphere bias for face processing (Dutta and Mandal, 2002, Hugdahl et al., 1993, De Renzi et al., 1994, Rhodes, 1993). However, Vaid and Singh (1989) have argued that the effect may not purely reflect a right hemisphere processing advantage but rather an interaction between a long practised directional scanning bias of the participants and cerebral lateralisation. Further evidence for this claim is provided by Heath et al. (2005), who reported increased leftward biases in the perception of facial affect amongst readers of left to right scripts compared to readers of right to left scripts or bi-directional readers. With regard to analysis of eye movements, partial support for the Vaid and Singh hypothesis comes from a recent study by Butler et al. (2005) in which a clear relationship between the chimeric face bias and the eye fixation pattern was reported: on trials where participants showed a left perceptual face bias they produced significantly more left saccades and fixated for longer on the left side of the chimeric face.

At the same time though, Ferber and Murray (2005) published a prism adaptation study disputing any clear link between chimeric face bias and eye-movement patterns: although they found a left eye movement bias with a left chimeric face bias, when eye movements were shifted to the right (using prism adaptation), they failed to observe a concurrent shift in the chimeric face judgements. From this finding, the authors argue that overt motor responses such as eye movements are not required in order to produce perceptual biases in chimeric face judgements.

Following their argument, if we were to exclude eye movements completely from the behaviour of our subjects, the perceptual bias should still be present. Relevant to this are the findings of David (1993), who employed stimuli that required an emotional judgement, and reported a left field perceptual bias with chimeric face drawings exposed for 120 ms.

In the present study we were asking if a reliable left perceptual bias could be demonstrated employing the realistic face stimuli used in Butler et al. (2005), as subjects may not engage in normal face processing mechanisms when stimuli look obviously chimeric (Burt and Perrett, 1997) or oversimplified as can be the case for line drawings (Luh et al., 1991). As in David's study, we presented stimuli below the threshold of eye movements yet requiring a gender judgement as again there is evidence that gender judgements reflect processes closer to face processing mechanisms such as identification, compared to emotional judgments (Bülthoff and Newell, 2004, Rossion, 2002).

Seventeen right-handed participants were exposed to blocks of single gender and chimeric face images, based on the set described in Butler et al. (2005). Stimuli were exposed for 100 ms and participants made gender decisions to each stimulus.

Section snippets

Results

Of the whole faces 26% of male and 18% of female faces were incorrectly identified. This was expected as the stimuli were androgynous looking faces (see Butler et al., 2005) and presented very briefly. A paired sample t-test of male and female accuracy indicated no significant difference (t(16) = 1.94, p = 0.07).

Interestingly, information on the left hand side of the face again dominated the gender decisions in that 55% of all responses were based on the left side of the chimeric face (t(16) = 3.56, p

Discussion

We have thus shown that even with a very brief exposure duration that should not allow any eye movements, it is still possible to obtain a significant left perceptual bias with gender based chimeric images. These findings of a perceptual bias at short presentation times are in line with David (1993). Moreover, the results can be taken as support for Ferber and Murray (2005), and also older studies by Grega et al. (1988) and Phillips and David (1997) who failed to find a link between

Experimental procedures

Seventeen right-handed participants1 (10 female and 7 male, mean age 21, S.D. = 2.1) took part in the study. All gave informed written consent and procedures were approved by the Glasgow University Ethics Review Board.

Stimuli were the same as described in Butler et al. (2005), although reduced to 8 bit format to satisfy the requirements of the Experimental Run

Acknowledgments

We would like to thank the Royal Society of Edinburgh/Lloyds TSB Foundation for supporting SHB, David I. Perrett and D. Michael Burt for the stimulus generation, and the most helpful comments of four anonymous reviewers on earlier versions of the manuscript.

References (30)

Cited by (0)

View full text