This research was supported by Economic and Social Research Council (Polaris House,
Swindon, England) doctoral studentship R42200134262 to Russell R. C. Hutter.

Address correspondence to Russell R. C. Hutter, School of Psychology and Clinical

Language Science, University of Reading, Earley Gate, Reading, RG6 6AL, UK; r.r.c.hutter@
reading.ac.uk (e-mail).

253

The Journal of Social Psychology, 2006, 146(2), 253–256

Copyright © 2006 Heldref Publications

Implications of Cognitive Busyness for the

Perception of Category Conjunctions

RUSSELL R. C. HUTTER

School of Psychology and Clinical Language Science

University of Reading, UK

RICHARD J. CRISP

School of Psychology

University of Birmingham, UK

SOCIAL CATEGORIZATION—in a constantly changing social world where such
classifications of others are becoming increasingly complex, one’s ability to deal
with inconsistencies is essential to maintaining coherent representations of them.
For instance, increasing multiculturalism means that ethnic identities no longer
map directly onto national identities. Occupational roles are becoming decreas-
ingly gender stereotypical. As categories combine in sometimes counterstereo-
typical and consequently surprising ways (female engineers, male nurses; see
Hastie, Schroeder, & Weber, 1990; Kunda, Miller, & Claire, 1990), perceivers
must process and incorporate information that is inconsistent with their expectan-
cies. It is well established that dealing with inconsistent information in impres-
sion formation is cognitively effortful (Brewer, 1988; Fiske & Neuberg, 1990),
but it is unclear now how such processes of impression formation interact when
one is dealing with surprising category combinations. The present study extend-
ed recent research on the composition of category conjunctions (Hutter & Crisp,
2005) to test whether and how the resource-consuming processes of resolving
inconsistency affect the impressions that one forms about people who are defined
by a surprising combination of social categorizations.

Kunda et al. (1990) found that when one perceives a surprising category com-

bination, participants generated emergent attributes, attributes associated with a
conjunction that are independent of the constituent categories from which the

combination was derived (see also Hastie et al., 1990; and more recently, Hutter
& Crisp, 2005). For instance, Kunda and colleagues use a “Harvard-educated car-
penter” as their example. One may perceive such a combination as nonmaterial-
istic by virtue of its surprising nature, but nonmaterialistic is a characteristic that
is not typically associated with either constituent category (Harvard educated or
carpenter). If a social category combination is surprising, then one combines the
categories’ definitions into one that is distinct from those of either of the con-
stituent categories. Surprising combinations are “more than the sum of their
parts.” In contrast, unsurprising combinations (e.g., Harvard-educated lawyer)
seem to be comprised simply of an additive combination of constituents, with no
emergent attributes.

As far as cognitive processing is concerned, one possibility that researchers

have raised is that surprising combinations trigger a process of inconsistency res-
olution. That is, their nature leads to heightened cognitive activity to resolve the
conflicting stereotypes, a result of which is a redefinition of the combination
using emergent attributes. In the broader literature on stereotyping, researchers
have defined inconsistency resolution well. It is the process by which one resolves
seemingly incongruent information (Hastie, 1981; Srull & Wyer, 1989). Badde-
ley (1996) and Macrae, Bodenhausen, Schloerscheidt, and Milne (1999) have
shown it to be a resource-depleting process. Therefore, if resources are scarce,
then the scarcity will inhibit one’s ability to resolve inconsistencies, including
those involved in dealing with category combinations. Although previous
researchers have suggested that the emergence of novel attributes that people use
to redefine surprising combinations may rest on a cognitively effortful inconsis-
tency-resolution process, that hypothesis remains empirically unproven. So, in
the present study, we tested the prediction that if participants were engaged in a
concurrent task that demanded cognitive resources, they would not generate
emergent attributes when also encountering a surprising combination. A concur-
rent task would reduce the available resources that are necessary for inconsisten-
cy resolution, and the scarcity of resources would be evident in fewer emergent
attributes than in a control condition.

The present experiment replicated a paradigm that Hutter and Crisp (2005)

previously used, but it had the addition of a cognitive-load manipulation. Partici-
pants were 39 undergraduate students whom we randomly allocated to a 2 (cog-
nitive load: high vs. low)

× 2 (attribute: emergent vs. constituent) mixed design

with repeated measures on the second factor. Participants provided informed con-
sent and had to generate up to a maximum of 20 attributes to describe an “Oxford-
educated bricklayer,” a surprising category combination, in precisely 120 s. Hut-
ter and Crisp empirically established that this combination was indeed “surprising”
to the same participants. In the present study, 19 participants received a task that
Macrae et al. (1999) established as resource-depleting: random-number genera-
tion. In the present study, we gave these instructions: “Whilst completing the above
please say aloud a number between 1 and 5 every second. Do not repeat the num-

254

The Journal of Social Psychology

ber consecutively.” In the control condition, 20 participants did not receive this
instruction. We compared attributes that participants generated with those of 40
other participants who had previously listed attributes for one of the two constituent
categories: “Oxford educated” or “bricklayer” (from Hutter & Crisp, Experiment
1). We made this comparison to test whether the attributes that participants gener-
ated (a) originated with either of these constituents or (b) were independent of the
constituents (i.e., emergent). We thanked and debriefed the participants.

To calculate the number of emergent attributes for the combined catego-

ry, we used the procedure outlined by Hastie et al. (1990; see also Hutter &
Crisp, 2005) whereby emergent attributes are defined as a concept used by per-
ceivers only when describing the category combination. In contrast, we defined
constituent attributes as attributes that were also generated by participants in
the comparison constituent categories. Two independent coders who were
blind to the experimental hypotheses performed this redundancy check using
a thesaurus to eliminate any words within each participant’s response list that
were repeated. This process resulted in (a) the removal of 24 attributes over-
all, leaving 214 attributes from the original 238 generated for Coder 1; and (b)
the removal of 14 attributes, leaving 224 attributes from the original 238 gen-
erated for Coder 2. The independent coders then classified the attributes as
either emergent or constituent. The interrater agreement on the classification
of attributes across coders was acceptable for both emergent attributes, r(39)
= .73, p < .0005, and constituent attributes, r(39) = .60, p < .0005. So, we took
the mean of the two coders’ ratings as a single index of generated emergent
and constituent attributes.

The 2 (load: high load vs. low load)

× 2 (attribute: emergent vs. con-

stituent) analysis of variance revealed a main effect of load, F(1, 37) = 5.05, p
< .05,

η

p

2

= .120 (M

low load

= 3.06, M

high load

= 2.24), but not of attribute, F(1,

37) = .133, p = .718. However, the predicted interaction between load and
attribute, F(1, 37) = 2.90, p < .10,

η

p

2

= .073, marginally qualified this main

effect. Planned comparisons revealed, in line with the hypothesis, significantly
more emergent attributes generated in the low-load condition than in the high-
load condition, t(37) = 2.57, p < .05 (M

low load

= 3.25, M

high load

= 1.95). In con-

trast, we found no difference for the number of constituent attributes generated
between the low-load condition and the high-load condition, t(37) = .846, p =
.403 (M

low load

= 2.88, M

high load

= 2.53).

These findings support the idea that a person’s generation of emergent—but

not constituent—attributes when encountering a surprising category combination
requires a cognitively demanding inconsistency resolution process. While cogni-
tive load inhibited the generation of emergent attributes, it had no effect on the
generation of constituent attributes. Having established the role of inconsistency
resolution in the perception of surprising category combinations, researchers need
to clarify the precise nature of this resource-demanding task and its implications
for social perception, stereotyping, and intergroup relations.

Hutter & Crisp

255

REFERENCES

Baddeley, A. (1996). Exploring the central executive. Quarterly Journal of Experimental

Psychology, 49A, 5–28.

Brewer, M. B. (1988). A dual process model of impression formation. In R. Wyer & T.

Srull (Eds.), Advances in social cognition (Vol. 1, pp. 1–36). Hillsdale, NJ: Erlbaum.

Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation from catego-

ry based to individuating process: Influences of information and motivation on atten-
tion and interpretation. Advances in Experimental Social Psychology, 23, 1–74.

Hastie, R. (1981). Schematic principles in human memory. In E. T. Higgins, C. P. Herman,

& M. P. Zanna (Eds.), Ontario Symposium: Vol. 1. Social cognition (pp. 39–88). Hills-
dale, NJ: Erlbaum.

Hastie, R., Schroeder, C., & Weber, R. (1990). Creating complex social conjunction cate-

gories from simple categories. Bulletin of the Psychonomic Society, 28, 242–247.

Hutter, R. R. C., & Crisp, R. J. (2005). The composition of category conjunctions. Per-

sonality and Social Psychology Bulletin, 31, 647–657.

Kunda, Z., Miller, D. T., & Claire, T. (1990). Combining social concepts: The role of causal

reasoning. Cognitive Science, 14, 551–577.

Macrae, C. N., Bodenhausen, G. V., Schloerscheidt, A. M., & Milne, A. B. (1999). Tales

of the unexpected: Executive function and person perception. Journal of Personality and
Social Psychology, 76, 200–213.

Srull, T. K., & Wyer, R. S. (1989). Person memory and judgment. Psychological Review,

96(1), 58–83.

Received January 17, 2005

Accepted December 13, 2005

256

The Journal of Social Psychology