Belief–desire reasoning in the explanation of behavior,Do actions speak louder than words

心理理论

Cognition 105 (2007) 184–194

/locate/COGNIT

Brief article

Belief–desire reasoning in the explanation of behavior: Do actions speak louder than words?

Annie E. Wertz, Tamsin C. German¤

Department of Psychology, University of California, Santa Barbara, CA 93106-9660, USAReceived 13 July 2006; accepted 8 August 2006

Abstract

The mechanisms underwriting our commonsense psychology, or ‘theory of mind’, havebeen extensively investigated via reasoning tasks that require participants to predict the actionof agents based on information about beliefs and desires. However, relatively few studies haveinvestigated the processes contributing to a central component of ‘theory of mind’ – our abilityto explain the action of agents in terms of underlying beliefs and desires. In two studies, wedemonstrate a novel phenomenon in adult belief–desire reasoning, capturing the folk notionthat ‘actions speak louder than words’. When story characters were described as searching inthe wrong place for a target object, adult subjects often endorsed mental state explanations ref-erencing a distracter object, but only when that object was approached. We discuss how thisphenomenon, alongside other reasoning “errors” (e.g., hindsight bias; the curse of knowledge)can be used to illuminate the architecture of domain speciWc belief–desire reasoning processes.© 2006 Elsevier B.V. All rights reserved.

Keywords:Theory of mind; Explanation; Attribution; Belief–desire reasoning

“If we could do that well with predicting the weather, no one would ever get hisfeet wet; and yet the etiology of the weather must surely be child’s play comparedwith the causes of behavior” (Fodor, 1987, p. 4).

*This manuscript was accepted under the editorship of Jacques Mehler.Corresponding author. Tel.: +1 805 893 5618; fax: +1 805 893 4303.

E-mail address: german@psych.ucsb.edu (T.C. German).

心理理论

A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194185

According to naïve belief–desire psychology (e.g., Davidson, 1963; Dennett, 1987;Fodor, 1987),1 agents’ actions are caused by two broad classes of mental folk con-structs: beliefs and desires. Davidson (1963) argued that actions are caused by reasons– pairings of two types of mental state (or ‘propositional attitude’). ‘Pro’ attitudes,such desires for objects, motivate action. ‘Cognitive’ attitudes constrain the details ofthat action: where or how it will take place. If we know a person has a reason for aparticular action (e.g., the desire for x and the belief that x is in location A), we canpredict the action. Conversely, knowing about an action (e.g., search at location A),allows the generation of reasons that might have caused it (e.g., desire for x and thebelief that x is located at A).

Research into our commonsense capacity to understand the actions of others viabelief–desire reasoning has been largely undertaken from a developmental perspec-tive (though see also Amodio & Frith, 2006; Saxe, Carey, & Kanwisher, 2004), focus-ing on the capacity to predict actions from information about desires and beliefs. In aclassic task (Baron-Cohen, Leslie, & Frith, 1985) preschool participants meet Sally,who leaves her marble in one location before departing, at which time Ann moves themarble to another location. On Sally’s return, the child must predict where she willlook for the marble. This task shows reliable improvement across the preschoolyears. Three-year-old children tend to predict that Sally will search in the locationactually containing the marble, while 4-year-olds more often predict that Sally willsearch where she thinks the marble is – in the now empty location (Wellman, Cross, &Watson, 2001).

Models of the processes contributing to belief–desire reasoning in action predic-tion have recently emerged (Friedman & Leslie, 2004a, 2004b; German & Hehman,2006; Leslie, German, & Polizzi, 2005; Leslie & Polizzi, 1998). However, despite thecentral importance of explanation of action in ‘theory of mind’, developmental inves-tigations of belief–desire explanation tasks have been few and far between (e.g., Bart-sch & Wellman, 1989; Moses & Flavell, 1990; Wimmer & Mayringer, 1998) andstudies of belief–desire explanation in adults, models of which must eventually con-strain all developmental theories, even rarer.

One framework for interpreting action prediction tasks is the two component the-ory of belief–desire reasoning proposed by Leslie and colleagues (e.g., Leslie, Fried-man, & German, 2004). According to this view, belief–desire reasoning is based inpart on a neurocognitive mechanism that takes as input information about thebehavior of social agents and generates candidate representations that might havecontributed to this behavior. A second mechanism has the task of selecting amongthe candidate mental states (see German & Hehman, 2006; Leslie etal., 2004; Yazdi,German, Defeyter, & Siegal, 2006, for more discussion). While extensive work hasassessed the nature of selection processing in prediction tasks (Friedman & Leslie,2004a, 2004b; Leslie etal., 2005), the idea that speciWc mental state representationsmight be spontaneously generated from information about behavior of social agents

The term ‘naïve’ is used here to distinguish everyday psychology from products of scientiWc psycholog-ical investigation. As Fodor (1987) argues, everyday psychology is anything but naïve in terms of itssuccess.1

心理理论

186A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194

has not been tested directly by researchers adopting this framework. However, anextensive literature in social psychology has shown that causal explanations (attribu-tions) of social behavior are often (i) focused on internal dispositions of actors (seeSmith, 1994, for a review), and that (ii) inferences about internal traits can be madespontaneously and unknowingly (Winter & Uleman, 1984).

In the experiments reported here, we extend this work to address how speciWcmental state representations (e.g., belief/desire representations) are generated inresponse to a speciWc aspect of social behavior. Adult participants received scenariosbased on the false belief task, extended to include the characters’ mistaken search.Two types of scenario diVered in terms of the mental state representations predictedto be generated in response to the search action. The presence of these diVerent men-tal state representations in the cognitive system was assessed via requiring subjects toendorse or reject candidate explanations for the search action.

In the experimental scenarios, the false belief scenario was altered by including asecond object (hereafter the ‘distracter’ object) at the location where the characterleaves her desired object (hereafter the ‘target’ object). When the character searchesfor the target object, her action is directed at a location that contains the distracterobject. The control scenarios were identical except that the distracter object was inthe second location, such that the character searches an empty location on her return.

We predicted that mental states referencing the distracter object would be gener-ated in the cognitive system in response to the search action, and would enter into theexplanation process in the experimental conditions only. In Experiment 1, we testedfor the presence mental states referencing desires about the distracter object. Thisshould manifest as errors in rejecting this kind of explanation, as compared withendorsing or rejecting other explanations consistent with the story.

Experiment 1

Method

Participants

Sixty Wve adults (41 females and 24 males, mean age 19) participated for classcredit. Participants were randomly assigned to either the ‘action toward distracterobject’ (ADO) condition (ND34) or ‘action toward empty location’ (AE) condition(ND31). The proportion of males to females in each condition was approximatelyequivalent.

Design and materials

Participants were presented with a series of 40 text-based reasoning tasks. Alltasks consisted of stories involving two agents, two objects and two locations. Eachshort story described a scenario in which character 1 places a target object in onelocation and then leaves. At this time character 2 moves the target object to a secondlocation. On her return, character 1 approaches the initial (wrong) location toretrieve the target object. For each task, the story was followed by one candidate

心理理论

A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194187

explanation for the mistaken search. Participants’ task was to decide as quickly aspossible whether the candidate explanation was a correct explanation for the searchaction. The conditions diVered only in the location of the distracter object.

In the ADO condition, the Wrst location also contained the distracter object. Thus,the agent’s search for the target object is directed at a location also containing thedistracter object. In the AE condition, the distracter object was in the second loca-tion. Therefore, the search was directed toward an empty location. An example storyfrom each condition is shown in Table 1.

Four types of candidate explanations were assessed in each condition. Eachexplained the action in terms of a mental state that character 1 might hold. Two ref-erenced the character’s desire: one for the target object, the other for the distracterobject. The other two explanations referenced the character’s belief about the targetobject: a correct (false) belief about the location of the target object and an incorrect(true) belief about the location of the target object. Participants saw just one explana-tion per story, such that each explanation type was assessed ten times. The candidateexplanations associated with the example story appear in Table 2.

Procedure

The 40 tasks were presented in random order via E-Prime (Psychology SoftwareTools, Inc.) software. Each task began with the presentation of the short story on thecomputer screen. Participants read the story at their own pace before pressing a keyto advance. Next, an explanation question was displayed for 2 seconds (e.g., “Whydoes Mary go there?”), followed immediately by a screen displaying one of the fourpossible candidate explanations. Participants were required to press one key toendorse the explanation or another to reject it. Responses and response latencieswere collected. Two practice trials preceded the 40 test trials.

Table 1

Example stories for Experiments 1 and 2

Condition

Action toward distracter object (ADO)

Mary puts her hairdryer next to her perfume in the

drawer and leaves the room. While Mary is in the

shower, Gina moves the hairdryer to the cabinet.

Mary comes back into the room for her hairdryer.

She goes directly to the drawer.Action toward empty location (AE)Mary puts her hairdryer in the drawer and leaves the room. While Mary is in the shower, Gina moves the hairdryer next to her perfume in the cabinet. Mary comes back into the room for her hairdryer. She goes directly to the drawer.

Note: Italics highlight the diVerence between the two conditions.

Table 2

Candidate explanations for Experiment 1

Explanation type

Desire for target (DT)¤

Desire for distracter (DD)

False belief about target (FBT)¤

True belief about target (TBT)¤Example explanationBecause she wanted to get her hairdryer from the drawerBecause she wanted to get her perfume from the drawerBecause she thought her hairdryer was in the drawerBecause she thought her hairdryer was in the cabinet‘Correct’ explanation for ADO and AE conditions.

心理理论

188A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194

Fig.1. Mean errors in performance for each candidate explanation according to condition for Experiment

1 (left panel) and Experiment 2 (right panel). For Experiment 1, explanation types were ‘desire for targetobject’ (DT; correct answerDendorse), ‘desire for the distracter object’ (DD; correct answerDreject),‘false belief about target object’ (FBT; correct answerDendorse) and ‘true belief about target object’(TBT; correct answerDreject). For Experiment 2 the explanation types were ‘belief held about targetobject’ (BHT; correct answerDendorse), ‘belief held about distracter object’ (BHD; correctanswerDreject), ‘belief not held about target object’ (BNHT; correct answerDreject) and ‘belief not heldabout distracter object’ (BNHD; correct answerDreject). All error bars represent §1 standard erroraround the mean.

Results

Responses consistent with the information presented in the story were scored ascorrect; responses inconsistent with that information were scored as errors. Thus, thecorrect responses were: endorsing the desire for the target object, endorsing a correct(false) belief about the target object location, rejecting the desire for the distracterobject, and rejecting an incorrect (true) belief about the target object location. Theconverse response for each explanation counted as an error.

The error scores for each explanation type, arranged by condition, appear in Fig.1(left panel). Error rates were low for all explanation types in both conditions (Merror range 4.2–13.9%), the exception being the rejection of the ‘desire for distracterobject’ explanation in the ADO condition (MD32.1%). This pattern was conWrmedin a 2 (condition; ADO vs. AE) by 4 (explanation type; ‘desire for target’ [DT], ‘desirefor distracter’ [DD], ‘false belief about target’ [FBT], ‘true belief about target’ [TBT])analysis of variance (ANOVA), with repeated measures for the ‘explanation type’factor.2 There was no overall eVect of condition but there was a main eVect ofexplanation type (F(3,189)D17.28, p<.0001, 2D.22), qualiWed by a condition by

The same 2£4 ANOVA was conducted on the response time data, with error trials and outliers (de-Wned as response latencies >3 SD above the mean) removed. This revealed a main eVect of explanationtype (F(3,189)D17.02, p<.0001, 2D.21) in which the endorsement of correct explanations was faster thanthe rejection of incorrect explanations.2

心理理论

A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194189

explanation type interaction (F(3,189)D5.01, pD.002, 2D.07).3 The interactionreXects a greater number of errors when rejecting the DD explanation in the ADOcondition compared to the AE condition.

This result was supported by independent samples t-tests for each type of explana-tion across conditions. Consistent with the ANOVA Wndings, only the error rate forthe DD explanations diVered across the AE and ADO conditions (t(63)D2.47, pD.02,dD.59). The elevated error rate for rejecting the DD explanations in the ADO condi-tion indicates that participants reliably endorsed this explanation as an acceptablereason for the story character’s action.

Discussion

Participants endorsed desire explanations for an object not speciWed as the targetof search (the distracter object) when the search was directed toward that object (asin the ADO condition). This suggests that a mental state representation specifying adesire for the distracter object was generated in response to information about thesearch and entered into the explanation process.

A question remains about the nature of the information generated by the belief–desire reasoning system when an agent approaches an object. One possibility is that thesystem only creates representations of the character’s possible goal – a desire for thedistracter object as shown in Experiment 1 – and no other mental state representationsaccompany it. An alternative is that something akin to a ‘reason’ in Davidson’s (1963)sense is created: a representation of the character’s (true) belief about the location ofthe distracter object may be produced alongside the representation of the desire. Thisidea is consistent with the proposal that belief and desire information are processed inparallel – a proposal driving some recent models of action prediction (Friedman &Leslie, 2004a, 2004b; Leslie etal., 2004, 2005). This idea is tested in Experiment 2.Experiment 2

Method

Participants

Seventy six adults (61 females and 15 males, mean age 19) participated for classcredit. Participants were randomly assigned to either the ADO (ND38) or AE condi-tion (ND38).

Design and materials

The design and materials were identical to Experiment 1, with the exception of thecandidate explanations assessed. Each of the four explanation types was framed as abelief. The object to which the belief referred varied (‘target’ or ‘distracter’ object) as3We report partial 2 throughout.

心理理论

190A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194

Table 3

Candidate explanations for Experiment 2

Explanation Type

Belief held about target (BHT)¤

Belief held about distracter (BHD)

Belief not held about target (BNHT)

Belief not held about distracter (BNHD)aExample ExplanationaBecause she thought her hairdryer was in the drawerBecause she thought her perfume was in the drawerBecause she thought her hairdryer was in the cabinetBecause she thought her perfume was in the cabinetExample explanations from ADO condition; the location of the objects in the BHD and BNHD expla-nations was inverted in the AE condition.

¤‘Correct’ explanation for ADO and AE conditions.

did the status of the belief (held by the character or not). The candidate explanationsused in Experiment 2 appear in Table 3.

Procedure

The procedure was identical to that used in Experiment 1.

Results

Responses consistent with the explicitly presented story information were scoredas correct. Thus, correct responses for Experiment 2 were: endorsement of a (false)belief explanation about the initial location of the target object, and rejections of theother three explanation types. All other responses were considered errors. The errorscores for each explanation type, arranged by condition, appear in Fig.1 (rightpanel).

The results show that errors rates were low for all explanation types in bothconditions (M error range 2.9–10.3%), except when participants were required toreject a belief about the distracter object in the approach location in the ADOcondition (MD42.4%). This was conWrmed in a 2 (condition; ADO vs. AE) by 4(explanation type; ‘belief held about target object’ [BHT], ‘belief held about dis-tracter object’ [BHD], ‘belief not held about target object’ [BNHT] and ‘belief notheld about distracter object’ [BNHD]) ANOVA, with repeated measures for the‘explanation type’ factor.4 There was a main eVect of condition (F(1,74)D10.98,pD.001, 2D.13) and a main eVect of explanation type (F(3,222)D23.43, p<.0001, 2D.24), qualiWed by a condition by explanation type interaction (F(3,222)D37.88,p<.0001, 2D.34). The interaction indicates elevated errors in response to the(true) belief held by the character about the location of the distracter object(BHD) in the ADO condition only.

This result was further supported by independent samples t-tests for each explana-tion type across condition. As indicated by the ANOVA, the error rates for the BHD

The same 2£4 ANOVA conducted on the response time data with error trials and outliers (deWned asabove) removed, revealed a main eVect of explanation type (F(3,222)D12.35, p<.0001, 2D.14). This pat-tern of results stems from endorsements of the correct explanation (the belief held by the character aboutthe target object) being faster than rejections of the other three explanations in both conditions.4

心理理论

A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194191

explanations diVered across the AE and ADO conditions (t(74)D6.27, p<.0001,dD1.17). The elevated error rate for rejecting the BHD explanations in the ADO con-dition means that participants reliably endorsed this type of explanation as an accept-able reason for the story character’s action. Error rates in both the ‘belief not heldabout target object’ (BNHT) and ‘belief not held about distracter object’ (BNHD)explanations also diVered across conditions (t(74)D2.56, pD.01, dD.57 andt(74)D2.09, pD.04, dD.49, respectively). However, note that overall error rates forthese explanations were low (MD3.7–10.3%) as was the eVect sizes of comparison tothe AE condition, as compared to the error rate (MD42.4%) and eVect size for theBHD explanation comparison.

Discussion

Experiment 2 shows that information about a social agent’s action provokes thebelief–desire reasoning system to generate belief representations that might have con-tributed to that behavior. When the action is directed toward an object not men-tioned as the target of search (the distracter object), participants nevertheless endorsea true belief about the location of that object as an explanation for the action. Thisresult, in conjunction with the results from Experiment 1, suggests that the belief–desire reasoning system generates a ‘reason’ when witnessing an agent move towardan object. SpeciWcally, this ‘reason’ consists of a desire for that object and a truebelief about its location.

General discussion

Belief–desire reasoning has been characterized as comprising at least two kinds ofcognitive mechanisms – a representational system that parses observable behavior interms of possible mental state descriptions and an executive selection system thatdecides among the candidate mental state representations when explaining or pre-dicting actions (Leslie etal., 2004).

The current experiments address a key prediction derived from the proposedfunctional role of the representational system using tasks that assess the ability toexplain social behavior. Because the system is automatic, we predicted that itshould calculate candidate mental state representations based on the actions ofagents – even if those representations are at odds with the explicitly stated cause ofthe action. The results demonstrate that the representational system generates can-didate belief–desire explanations (‘reasons’) when presented with informationabout the actions of a social agent. When the actions of an agent are directedtoward a plausible goal object, albeit one that is irrelevant to the scenariodescribed, mental states related to that object are represented and inXuence deci-sions about what likely caused the observed action. SpeciWcally, a desire for theobject and a true belief about its location are calculated. When equivalent situa-tions containing the same agents, locations and objects are used, but when the

心理理论

192A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194

action is directed instead at an empty location, mental states referencing that irrel-evant object do not inXuence the explanation process.

We have characterized the mental state representations about the distracter objectas being generated when the agents’ actions were described. However, it is possiblethat the representations were calculated when the explanation probes appeared.Though the current studies cannot resolve this issue, evidence using an explanationproduction paradigm suggests that 3–5-year-old children and adults will explain asearch action in terms of mental states referencing approached distracter objects(German, 1995; Wertz, Levinger, & German, 2006, in preparation).

The notion that folk psychology includes attribution of mental states to agentsis neither unknown nor new (Heider, 1958; Heider & Simmel, 1944). In fact, theWeld of attribution research takes this notion as a starting point and focuses onthe more general dynamics of all agent-centered attributions (e.g., the fundamen-tal attribution error; Jones & Davis, 1965; Smith, 1994; spontaneous trait infer-ences; Winter & Uleman, 1984). Our Wndings address how mental states—aspeciWc class of agent-centered attributions – are generated in response toobserved units of behavior.

The current Wndings are also complimentary to evidence for infants’ early reason-ing about agents in terms of the principle of ‘rational action’ (Gergely, Nádasdy, Csi-bra, & Bíró, 1995). A central tenet of this theory is that infants attribute goal statesinsituations where observable patterns of motions and their end states conform toprinciples of rationality, given the observed layout of obstacles in the world.Repeated motion toward an identiWed end-state is a critical determinant of theinfants’ decision about whether that entity is goal-directed.

The phenomenon we report here of ‘errors’ in adult reasoning during explanationof behavior is consistent with research identifying biases that inXuence adult ‘theoryof mind’ performance, such as the ‘hindsight bias’ (Bernstein, Atance, Loftus, &MeltzoV, 2004; FischhoV, 1975), the ‘curse of knowledge’ (Birch & Bloom, 2004) andlimits on the eVective deployment of ‘theory of mind’ (Keysar, Shuhong, & Barr,2003). However, biases such as these should not be thought of as errors in the sensethat they might be considered to reveal ‘sub-optimal’ performance. Instead, theyshould be treated as windows into the design and functioning of the cognitive mecha-nisms that underlie our capacity to reason about mental states. In ecologically validcontexts it makes sense to consider a social agent’s behavior as testament to theirunderlying beliefs and desires, in addition to the assumptions or assertions aboutthose states available prior to the event. Indeed, the current evidence suggests that thefolk principle – ‘actions speak louder than words’ – is so sensible as to have beenincorporated into our cognitive architecture.

Acknowledgements

Thanks to April Blazer, Kimberly Burke, Emily Durrant, and Katherine Levingerfor assistance in conducting these experiments and members of the Cognition andDevelopment Laboratory and the Center for Evolutionary Psychology at UCSB for

心理理论

A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194193

helpful discussions. Special thanks to David Pietraszewski and Nancy Collins forcomments on a previous draft of this manuscript. T.P.G. was supported by a grantfrom the UCSB Academic Senate.

References

Amodio, D. M., & Frith, C. D. (2006). Meeting of minds: the medial frontal cortex and social cognition.

Nature Reviews, Neuroscience, 7, 268–277.

Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a “theory of mind”? Cogni-

tion, 21, 37–46.

Bartsch, K., & Wellman, H. M. (1989). Young children’s attribution of action to beliefs and desires. Child

Development, 60, 946–964.

Bernstein, D. M., Atance, C., Loftus, G. R., & MeltzoV, A. (2004). We saw it all along: visual hindsight bias

in children and adults. Psychological Science, 15, 264–267.

Birch, S. A. J., & Bloom, P. (2004). Understanding children’s and adults’ limitations in mental state reason-

ing. Trends in Cognitive Sciences, 8, 255–260.

Davidson, D. (1963). Actions, reasons and causes. Journal of Philosophy, 60, 685–700.

Dennett, D. (1987). The intentional stance. MA: MIT Press.

FischhoV, B. (1975). Hindsight does not equal foresight: the eVect of outcome knowledge on judgment under

uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1, 288–299.Fodor, J. A. (1987). Psychosemantics. Cambridge, MA: MIT Press..

Friedman, O., & Leslie, A. M. (2004a). Mechanisms of belief–desire reasoning: Inhibition and bias. Psycho-

logical Science, 15, 547–552.

Friedman, O., & Leslie, A. M. (2004b). A developmental shift in processes underlying successful belief–

desire reasoning. Cognitive Science, 28, 963–977.

Gergely, G., Nádasdy, Z., Csibra, G., & Bíró, S. (1995). Taking the intentional stance at 12 months of age.

Cognition, 56, 165–193.

German, T.P. (1995). Children’s explanations of action: desires versus beliefs in ‘theory of mind’. Unpub-

lished Ph.D. thesis, University of London.

German, T. P., & Hehman, J. A. (2006). Representational and executive selection resources in ‘theory of

mind’: evidence from compromised belief–desire reasoning in old age. Cognition, 101, 129–152.Heider, F. (1958). The psychology of interpersonal relations. New York: John Wiley & Sons Inc..

Heider, F., & Simmel, M. (1944). An experimental study of apparent behavior. American Journal of Psy-

chology, 57, 243–259.

Jones, E. E., & Davis, K. E. (1965). From acts to dispositions: the attribution process in person perception.

In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 2, pp. 220–266). New York:Academic Press.

Keysar, B., Shuhong, L., & Barr, D. J. (2003). Limits on theory of mind use in adults. Cognition, 89, 25–41.Leslie, A. M., Friedman, O., & German, T. P. (2004). Core mechanisms in ‘theory of mind’. Trends in Cog-

nitive Sciences, 8, 529–533.

Leslie, A. M., German, T. P., & Polizzi, P. (2005). Belief–desire reasoning as a process of selection. Cognitive

Psychology, 50, 45–85.

Leslie, A. M., & Polizzi, P. (1998). Inhibitory processing in the false belief task: two conjectures. Develop-

mental Science, 1, 247–253.

Moses, L. J., & Flavell, J. H. (1990). Inferring false beliefs from actions and reactions. Child Development,

61, 929–945.

Saxe, R., Carey, S., & Kanwisher, N. (2004). Understanding other minds: linking developmental psychol-

ogy and functional neuroimaging. Annual Review of Psychology, 55, 87–124.

Smith, E. R. (1994). Social cognition contributions to attribution theory and research. In P. G. Devine, D.

L. Hamilton, & T. M. Ostrom (Eds.), Social cognition: Impact on social psychology (pp. 77–108). SanDiego, CA: Academic Press.

心理理论

194A.E. Wertz, T.C. German / Cognition 105 (2007) 184–194

Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theory mind development: the truth

about false-belief. Child Development, 72, 655–684.

Wertz, A. E., Levinger, K., & German, T.P. (2006). University of California, Santa Barbara, Manuscript in

preparation.

Wimmer, H., & Mayringer, H. (1998). False belief understanding in young children: explanations do not

develop before predictions. International Journal of Behavioral Development, 22, 403–422.

Winter, L., & Uleman, J. S. (1984). When are social judgments made? Evidence for the spontaneousness of

trait inferences. Journal of Personality and Social Psychology, 47, 237–252.

Yazdi, A. A., German, T. P., Defeyter, M. A., & Siegal, M. (2006). Competence and performance in belief–

desire reasoning across two cultures: the truth, the whole truth and nothing but the truth about falsebelief? Cognition, 100, 343–368.

本文来源：https://www.bwwdw.com/article/f7hj.html