The efficiency of binding spontaneous trait inferences to actors’ faces

https://doi.org/10.1016/S0022-1031(03)00059-3Get rights and content

Abstract

Three experiments tested the hypothesis that the process of binding spontaneous trait inferences (STIs) to actors’ representations is relatively independent of attentional resources. Participants were presented with faces paired with single behaviors. Binding of STIs to actors was revealed by a higher false recognition of implied traits paired with actors’ faces than of implied traits randomly paired with other familiar faces. This effect replicated when each face–behavior pair was presented for 2 s (Experiment 1), when the processing of the information was shallow (Experiment 2), and when participants performed a secondary task during the presentation of behaviors (Experiment 3). Experiment 4 showed that explicit on-line trait judgments of the actor, but not explicit behavior judgments, predicted the false recognition of implied traits in the context of the actor. The possibility that the process of binding STIs to actors’ representations is automatic is discussed.

Introduction

Imagine yourself out for a midnight stroll in the city, occasionally sighting strangers doing trait-diagnostic things. This one up ahead mutters to himself and greets lampposts politely. That one whistles a sprightly tune to herself. Without even realizing it, you make inferences about each (mad and cheerful), and stroll on. Contrast this with a midday dash to a lunch appointment when you are preoccupied with what you are going to say when you get there. You pass hundreds of strangers and notice most of them, including these two. One mutters to himself as before, and another whistles a tune, as before. Assuming that you infer mad and cheerful as before, two questions arise. First, have you made inferences about the people or merely about their behavior? Is the man mad, or is he just behaving like a madman? Is she a cheerful person, or just behaving cheerfully, perhaps only for the moment? Second, will you remember who did (or was) what? In your dash to lunch, will you remember which person among all those strangers was cheerful, or merely that someone was cheerful? Keeping straight who did (or was) what seems much more likely on your midnight stroll, when you were not preoccupied and had fewer strangers to remember. Will you keep them straight at midday?

We can restate these two questions more precisely. First, are spontaneous trait inferences (STIs) from people’s behavior merely about the behavior, or are they also about the actor performing the behavior? Second, how efficiently are such trait inferences linked or bound to particular appropriate actors? The first three studies in this paper address this second question, using a new false recognition paradigm introduced by Todorov and Uleman (2002). The last study addresses the first question.

Many experiments have shown that simply reading about or observing a single behavior by another person is sufficient to trigger a trait inference, even when the observer does not intend to make an inference or to form an impression of the actor (for a review see Uleman, Newman, & Moskowitz, 1996). That is, people make spontaneous trait inferences. Based on this evidence, it has been commonly assumed that spontaneous behavior-to-trait inferences are an automatic process (Bargh, 1994), but that these inferences are limited to the meaning of the behavior and do not have further implications for the actor (Bargh, 1994; Bassili, 1989; Park, 1989; Wyer & Srull, 1989). In other words, such unintended inferences result in higher accessibility of trait concepts in memory but are not linked to the representation of the actor.

However, two recent experimental paradigms have provided convergent evidence strongly suggesting that STIs are about the actor. Both the savings and the false recognition paradigms use faces as critical stimuli. In the savings paradigm (Carlston & Skowronski, 1994; Carlston, Skowronski, & Sparks, 1995), participants are presented with faces paired with behavioral paragraphs implying traits. Later they are asked to learn face–trait pairs. Participants are faster at learning faces paired with traits that were implied in the paragraph presented with the face, than faces paired with new traits, or paired with traits implied in a paragraph about a different face. This facilitation of learning suggests that the trait was previously inferred and linked to the actor’s face.

In the false recognition paradigm (Todorov & Uleman, 2002), participants are presented with faces paired with trait implying behavioral sentences (e.g., “Mary solved the mystery half-way through the book”). Later in a recognition test, participants are presented with the faces paired with trait words (e.g., “clever”) and asked to decide whether they saw the word in the sentence presented with the face. Participants are more likely to falsely recognize implied traits when these traits are paired with actors’ faces than when they are paired with other familiar faces presented earlier in the experiment.1

The evidence from the false recognition and savings paradigms suggests that STIs are linked to representations of the actors. However, none of the previous studies using these two paradigms has addressed the automaticity of these inferences. For example, it is possible that behavior-to-trait inferences are automatic (e.g., Bargh, 1994; Lupfer, Clark, & Hutcherson, 1990; Winter, Uleman, & Cunniff, 1985), but that the process of binding these inferences to actor representations depends on the availability of cognitive resources and could be easily blocked by manipulations that interfere with the processing of the behavioral information. The objective of the present research was to address the automaticity of binding STIs to actors.

Although the previous studies in the savings and false recognition paradigms have not addressed the automaticity of actor-linked STIs directly, several experiments have shown that these inferences are fairly robust. For example, the savings effect (faster learning of face and implied-trait pairs) is unaffected by encoding instructions (Carlston & Skowronski, 1994). This effect seems to be the same for familiarization, memory, and impression formation instructions. In the false recognition paradigm, Todorov and Uleman (2002, Experiment 4) have arrived at similar conclusions, although they showed that memory and impression instructions trigger different processes. Moreover, even under conditions of information overload when participants were presented with 120 unfamiliar faces, each paired with a different behavior, participants linked STIs to the actor representation (Todorov & Uleman, 2002, Experiments 5 & 6).

In both paradigms, actor-linked STIs are not dependent on behavior retrieval. In the false recognition paradigm, the false recognition of implied traits occurs even if participants do not recall or do not recognize the behavior (Todorov & Uleman, 2002, Experiments 5 & 6). Similarly, the savings effect occurs even if participants do not recognize the behavioral paragraphs and the learning task is a week later than the presentation of faces and paragraphs (Carlston & Skowronski, 1994; Carlston et al., 1995).

However, in all previous studies the presentation time of faces paired with behavioral information has been sufficiently long (ranging from 5 to 10 s) to allow participants to rehearse the information or develop some consistent strategy for memorizing the information. Also, during this initial presentation participants’ only task was to memorize or familiarize themselves with the information. None of the previous experiments used a simultaneous secondary task to tax participants’ cognitive resources, or a low-level processing goal that produces shallow processing. Thus, it is not clear whether actor-linked inferences occur when cognitive resources are severely constrained.

The first three experiments used the false recognition paradigm (Todorov & Uleman, 2002) to examine the efficiency of STI-actor binding. Efficiency is one characteristic of automatic processes (Bargh, 1994) and includes relative independence of attentional resources. Manipulations that deplete attentional resources include rapid presentation of stimuli (e.g., Bargh & Thein, 1985) and a concurrent cognitive load (e.g., Gilbert & Osborne, 1989). To the extent that a process is robust with respect to such manipulations, the process is efficient. Experiment 1 reported below used rapid pace presentation of face–behavior stimuli, and Experiment 3 put participants under a cognitive load, rehearsing 6-digit numbers. A process can also be efficient in the sense that it occurs even under shallow processing goals (Craik & Lockhart, 1972). In Experiment 2, participants were told that the experiment was about how people process grammatical information in the context of faces and were asked to count the nouns in each sentence.

All three manipulations—rapid pace, cognitive load, and shallow processing—should lead to less accurate recognition performance. Compared with control conditions, participants should be both more likely to falsely recognize traits (whether or not these traits are paired with the actor’s face) and less likely to correctly recognize presented traits. More importantly, if binding of STIs to actor representations is relatively independent of attentional resources, participants should be more likely to falsely recognize implied traits paired with actors’ faces than implied traits randomly paired with familiar faces, even under these conditions.

The evidence from the false recognition and savings paradigms clearly shows that people engage in spontaneous inferences and that these inferences are linked to the actor representations. However, the nature of these inferences is ambiguous. We (Todorov & Uleman, 2002) have assumed that these are trait inferences about the actor. However, a simpler assumption is that these are inferences about the meaning of the behavior that are merely associated with actor representations. Indeed, the false recognition and the savings findings do not discriminate between these two assumptions.

Experiment 4 was designed to test whether the false recognition of implied traits reflects spontaneous trait inferences about the actor rather than merely behavioral categorizations associated with the actor representations. We measured on-line trait judgments of the actor (“Is Robert a honest person?”) and on-line behavior judgments (“Is this a honest behavior?”). If spontaneous inferences are inferences about the actor, explicit on-line judgments of the actor, but not on-line behavior judgments, should predict the probability of false recognition of implied traits.

Section snippets

Experiment 1

In all previous experiments using the false recognition paradigm, each face–behavior pair was presented either for 10 s or for 5 s. In this experiment, the presentation time was reduced to 2 s. This, coupled with the relatively large number of face–behavior pairs (60), reduces the probability of rehearsing the behaviors and other deliberate strategies to memorize the information. The experiment also included a condition in which participants self-paced the presentation of face–behavior pairs,

Experiment 2

Experiment 2 manipulated the encoding goals of participants. They were either given memory instructions as in Experiment 1 or were asked to count the nouns in each sentence, ostensibly as part of a task on grammatical processing in the context of faces. Because the latter instruction does not require elaborating on the behavioral meaning or processing the actor’s face, it should lead to shallow information processing. In explicit memory tasks, where participants are asked to recall or recognize

Experiment 3

The first two experiments showed that actor-linked STIs occurred when the presentation pace of face-behavior stimuli was rapid and when the behavioral information was processed shallowly. Experiment 3 used an additional manipulation intended to interfere with strategic or controlled but not automatic processing. Participants processed the face–behavior stimuli under cognitive load. Each trial consisted of a presentation of a 6-digit number followed by the presentation of a face–behavior pair

Experiment 4

Are STIs about the actor, or merely about the behavior and linked to the actor? Are people inferring that Mary is a clever person, or merely did something clever? If STIs are about the actor rather than merely about the behaviors, then measures of person inferences, but not of behavior inferences, should predict the probability of false recognition of implied traits. Todorov and Uleman (2002) provided evidence that explicit person judgments on the implied traits predicted the false recognition

General discussion

If participants associate an inferred trait with the actor, they should be more likely to falsely recognize this trait in the context of the actor than in a different context. That was the case across the three experiments reported here and across the six experiments reported in Todorov and Uleman (2002). In these experiments, participants were more likely to falsely recognize implied traits paired with actors’ faces than implied traits randomly paired with familiar faces.

More important, the

Conclusions

Even if we do not intend to make a person-judgment or to form a person impression, but simply attend to another person’s behavior, this is sufficient to trigger inferences about the person. These inferences occur even if the behavior presentation is very brief, if we do not attend carefully to the meaning of the behavior, and if our attention is severely constrained by a secondary task. These inferences are about the person and have specific implications for our representation of that person.

References (31)

  • F.I.M. Craik et al.

    Levels of processing: A framework for memory research

    Journal of Verbal Learning and Verbal Behavior

    (1972)
  • L.L. Jacoby

    A process dissociation framework: Separating automatic from intentional uses of memory

    Journal of Memory and Language

    (1991)
  • J.S. Uleman et al.

    People as flexible interpreters: evidence and issues from spontaneous trait inference

  • J.A. Bargh

    Auto-motives: Preconscious determinants of social interaction

  • J.A. Bargh

    The four horsemen of automaticity: Intention, awareness, efficiency, and control as separate issues

  • J.A. Bargh et al.

    Automaticity in action: The unconscious as repository of chronic motives and goals

  • J.A. Bargh et al.

    Individual construct accessibility, person memory, and the recall-judgment link: The case of information overload

    Journal of Personality and Social Psychology

    (1985)
  • J.N. Bassili

    Traits as action categories versus traits as person attributes in social cognition

  • D.E. Carlston et al.

    Saving in the relearning of trait information as evidence for spontaneous inference generation

    Journal of Personality and Social Psychology

    (1994)
  • D.E. Carlston et al.

    Savings in relearning: II. On the formation of behavior-based trait associations and inferences

    Journal of Personality and Social Psychology

    (1995)
  • F.I.M. Craik et al.

    Depth of processing and the retention of words in episodic memory

    Journal of Experimental Psychology: General

    (1975)
  • W.N. Francis et al.

    Frequency analysis of English usage: Lexicon and grammar

    (1982)
  • D.T. Gilbert et al.

    The correspondence bias

    Psychological Bulletin

    (1995)
  • D.T. Gilbert et al.

    Thinking backward: Some curable and incurable consequences of cognitive busyness

    Journal of Personality and Social Psychology

    (1989)
  • L.L. Jacoby et al.

    Separating conscious and unconscious influences of memory: Attention, awareness, and control

    Journal of Experimental Psychology: General

    (1993)
  • Cited by (203)

    • Simplifying social learning

      2024, Trends in Cognitive Sciences
    • Nonverbal communication

      2023, Encyclopedia of Mental Health, Third Edition: Volume 1-3
    • Assessing the speed and spontaneity of racial bias in pain perception

      2022, Journal of Experimental Social Psychology
    View all citing articles on Scopus
    View full text