Supplemental Normal and Mirrored Game Block Analysis, Figure S1

Running head: CHESS MASTERS SHOW A HALLMARK OF FACE PROCESSING 1

Supplemental – Normal and Mirrored Game Block Analysis, Figure S1

We predicted that processing chess stimuli that demanded more attention might magnify tradeoff effects. To test this, we reversed the game pieces about the vertical midline on half of the chess displays (see Figure S1a). The continuous task contained 8 blocks of normal displays and 8 with this “mirror” transformation, known to impair recall memory among expert chess players (Gobet & Simon, 1996), though it does not make the displays illegal in chess nor does it eliminate skilled players’ recall advantage over less skilled players. Our thinking was that mirror-reversal would increase the capacity demands on chessboard recognition among chess experts, and we wanted to learn whether the tradeoff between the congruency effect with chess and the congruency effect with faces would be more pronounced in trial blocks with mirrored chess displays than trial blocks with normal displays.

Our original ANOVA of d’ scores included the within-group variable of presentation block (normal- vs. reversed-chessboard blocks), and it revealed the following interactions: block × stimulus, F(1,66) = 4.17, p =.045, 2=.002, block × stimulus × congruency F(1,66) = 5.30, p =.025, 2=.002, and a marginal block × expertise × congruency interaction, F(2,66) = 3.06, p =.05, 2=.003.[1] These interactions reflected the fact that the congruency effects were reduced in the mirror-reversed blocks, specifically with faces and specifically among the chess experts and players, as shown in Figure S1c. Note that the reduced face congruency effect shown by chess experts and players as compared to novices was largely restricted to the mirror-reversed blocks.



A related observation is that the congruency effect with faces was stronger in the unaltered blocks than in the mirror-reversed blocks among both experts (Ms = 1.06 and .72, respectively, SDs = .62 and .83), t(26) = 2.19, p = .04, and recreational players (Ms = 1.09 and .58, respectively, SDs = .68 and .94), t(21) = 2.76, p = .01), but not novices (Ms = 1.29 and 1.35, respectively, SDs = .62 and .83), t(19) = .39, p = .70. A conceptually similar association between expertise and a “face interference index” was reported by Gauthier, Curran, Curby & Collins (2003), adding support to the idea that common processes are involved in selective-attention failures in face and chess recognition among chess players.

Contrary to expectations, the congruency effect with chessboards did not differ reliably in mirror-reversed versus unaltered trial blocks in any participant group. We can only speculate on the reasons for this, but the overall accuracy of chessboard recognition (collapsing over congruent and incongruent trials) was not lower for mirror-reversed displays than for unaltered displays in any group. Instead, the trend was (non-significantly) in the opposite direction among experts (M d’s = 1.85 and 1.70, respectively), and was essentially 0 in the other two groups (M d’s = .97 and .99, respectively, for players and .79 and .71, respectively, for novices). This outcome contrasts with the deleterious effect of mirror-reversal on recall of chess displays (Gobet & Simon, 1996), and may suggest that while chess experts may find mirror-reversed displays more difficult to encode, they perceive them as more distinctive than unaltered displays. Such an effect would work against any tendency for performance detriments with mirror-reversed displays in recognition tests (where distinctiveness is an important factor). More research is needed on this point.

Supplemental References

Gauthier, I., Curran, T., Curby, K., & Collins, D. (2003). Perceptual interference

supports a non-modular account of face processing. Nature Neuroscience, 6(4), 428-432.

Gobet, F. & Simon, H. (1996). Recall of random and distorted chess positions:

Implications for the theory of expertise. Memory & Cognition, 24(4), 493-503.

[1] An ANOVA performed on the congruent-incongruent differences in d’ supported a reliable trial block × stimulus class interaction (precisely corresponding to the block × stimulus × congruency interaction in the original ANOVA) and a reliable block × expertise interaction (precisely corresponding to the block × expertise × congruency interaction in the original ANOVA).