1Recently, Beauvois & Dubois (2000; see also Dubois & Beauvois, 2001) have published a comprehensive synthesis of their contributions concerning the role of traits in lay personology. A central issue concerns the nature of the evaluative "good-bad" component of the meaning of traits. A main merit of Beauvois and Dubois' contributions is that they have completed the traditional "individual" approach of this problem with a typically "social" approach.
2This approach (e.g., Peeters, 1999a) deals with traits as intrinsic properties of the perceived target person. They differ from physical properties such as hair colour only in that they cannot be directly observed. However, their presence in a target person is inferred from the target's behaviour (abbreviated: TB). The trait is conceived as a covert but real disposition underlying the overt TB. This connection of the trait with the TB constitutes the nonevaluative or descriptive component of the trait meaning. For instance the descriptive meaning of "honest" refers to TBs such as "(S)he does not cheat". The evaluative meaning of the trait consists of positive (versus negative) affect that is attached to the trait's descriptive meaning and it reflects the (positive versus negative) desirability of the target's TB for the perceiver.
3One problem is that the evaluative meaning of traits such as honest is assumed to be a fixed semantic feature of the trait, while individual perceivers' desires can be idiosyncratic. In this way the honesty of a customs officer can be very negative for a smuggler. This problem has been met by assuming that evaluative trait meanings reflect the adaptive value of the trait for humans in general. Peeters (1983; see also: Beauvois, Dubois & Peeters, 1999) has demonstrated that this adaptive value can be defined from two general perspectives. The first perspective is that of the self, or equivalently, that of the possessor of the trait. Traits such as industrious and competent (versus their opposites) are expected to involve unconditionally positive (versus negative) adaptive consequences for the industrious and competent (versus lazy and incompetent) person him- or her-self. These traits define a power-related evaluative dimension referred to as (positive versus negative) self-profitability (abbreviated: SP). The second general perspective is that of the other who has to deal with the possessor of the trait. It defines a likeability-related dimension marked by traits such as generous and tolerant (versus their opposites) that are expected to involve unconditionally positive (versus negative) adaptive consequences for the other. This dimension has been referred to as "other-profitability" (abbreviated: OP). SP and OP are presumably universal dimensions akin to a wide variety of two-dimensional models of implicit personality theory reviewed by, a.o., Beauvois et al. (1999) and Vonk (1993). For instance, they correspond to the intellectual and social good-bad dimensions that, according to Rodenberg and colleagues, account for classic findings in impression formation such as Asch’s (1976) warm-cold effects (e.g., Rosenberg and Sedlak, 1972).
4Beauvois and Dubois have proceeded from the idea that traits are not only dispositions but also affordances. This means that the primary meaning content of a trait comprises more than the reference to TBs that constitute the trait's descriptive meaning. It includes a reference to the behaviour (or behavioural attitudes) others can or must engage in when dealing with the possessor of the trait ("other's behaviour or OB). For instance, the meaning of "honest" does not only refer to TBs such as "(S)he doesn't cheat" but also to OBs such as "S(h)e's a person others can trust". Whether a trait is evaluated as either a "good" or a "bad" property would depend on the trait’s affordances. It follows that the trait’s OBs constitute the trait's evaluative meaning.
5OBs can be idiosyncratic. For instance, a most banal attribute of one's sweetheart such as a birthmark can give occasion to playful expressions of tenderness that are highly personal and enacted by only one single lover. For this lover the birthmark represents specific affordances that make of it a most positively valued feature of his sweetheart. Similar idiosyncratic affordances, and the related OBs, constitute evaluations referred to as "personal desirabilities". They are distinguished from evaluations connected with widely acknowledged affordances and related OBs that constitute evaluations referred to as "social utilities" (Beauvois et al., 1999). Common evaluative trait meaning reflects social utility rather than personal desirability. Beauvois & Dubois (2000) have reviewed ample evidence that (a) the more a trait is categorized by judges as "evaluative" the more the trait is associated in perceivers' minds with OBs, (b) there's agreement among perceivers about which OBs belong to which traits, and (c) the OBs are at least as tightly associated with the traits as are the TBs that are assumed to underlie the traits' descriptive meanings. In this way, the trait "honest" is as tightly associated with OBs such as "is trusted by others" as with TBs such as "does not cheat others".
6Presumably the individual and social approaches are not incompatible but highlight different sides of the same coin. For instance, while the individual approach has highlighted the possible role of affect as a constituent of evaluative trait meaning, the social approach has highlighted the role of OBs. The present research resulted from an attempt to bridge the gap between these approaches making that they could profit from each other's achievements. Our starting point was the observation that the concept of OB, that is a hallmark of the social approach, is not completely alien to the individual approach. Affects that, according to the individual approach, constitute evaluative meaning, are basically emotional responses. Since Wundt (1896) there has been a tradition to associate the valences of these emotional-affective responses with behavioural approach-avoidance tendencies. This means that positive and negative evaluative meanings are conceived as incentive values for approach-avoidance behaviour. Approach and avoidance represent in point of fact general OB categories (the possessor of the positive or negative trait is approached or avoided by others, not by him- or her-self). In this way "approach" encompasses any OB that involves some interaction with a target that is afforded by that target. For instance, buying and smoking are possible approach acts relative to the object "cigarette". "Avoidance" encompasses any OB directed to the prevention or elimination of interactions with the target, e.g.: throwing away one's cigarettes.
7Meanwhile, Wentura, Rothermund, & Bak (2000) have demonstrated that evaluative trait meanings elicit very concrete motoric approach-avoidance tendencies. In one of their experiments participants were instructed to respond whenever a trait appeared on a screen. One group of participants responded by pressing (i.e., "approach") a response key, another group by withdrawing from (i.e., "avoid") the key. Positive traits were found to facilitate more the approach response and negative traits more the avoidance response. It may be questioned, of course, whether these elementary motoric approach-avoidance responses can be generalized to complex social OBs such as dating or giving up a girl. Nevertheless the results of Wentura et al. support the idea that the approach-avoidance concept captures some fundamental aspect of evaluation. Considering the affinity of that concept with the OB, it may provide an appropriate avenue towards the integration of acquisitions from the social and individual approaches. For instance, one could expect OBs to vary along an approach-avoidance dimension that reflects the evaluative dimension along which trait meanings vary, making that positive traits elicit approach-related OBs and negative traits elicit avoidance-related OBs.
8At this point, however, certain acquisitions of the individual approach suggest some amendations. Specifically Peeters & Czapinski (1990) have questioned the generality of the approach-avoidance interpretation of evaluation. They argued that a target person is permanently in his or her own presence making that the person can neither approach nor avoid him- or her-self. Only an "other" person can do that. Hence they concluded that the approach-avoidance interpretation was limited to the OP dimension of evaluation and could not be extended to the SP dimension. Consistent with that conclusion, Wentura et al. (2000) found effectively that the above-mentioned effects of trait valences on motoric approach and avoidance responses were produced exclusively by OP valences and not at all by SP valences. Generalizing to the OBs of Beauvois and Dubois, we can formulate the following hypothesis:
9The concrete OBs associated with evaluative trait meanings vary along an approach-avoidance dimension whereby the approach and avoidance values of the OBs reflect the positive respectively negative OP values of the traits rather than the traits' SP values.
10At a first glance, this hypothesis is contradicted by the tradition to associate approach-avoidance with "evaluation in general", which means: evaluation conceived as a single encompassing good-bad dimension. However, this tradition may have evolved from the fact that general evaluative (good-bad) ratings are usually made from the perspective of the other rather than from the perspective of the evaluated target him- or her-self. It follows that the usual evaluative ratings reflect the approach-avoidance related OP dimension rather than the SP dimension (Vonk, 1999).
11In addition, genuine SP values of traits can endorse OP value (and the correspondent approach-avoidance value) by interaction with the context (Peeters, 1992). For instance, positive OP value is accorded to a target's high competence (positive SP) if the target is perceived as a friend, while negative OP value is accorded if the target is perceived as an enemy. In this way stereotypes of the own group and of friendly groups have been found to be marked by positive SP rather than by positive OP (Peeters, 1993; Phalet and Poppe, 1997). These observations fit fairly well Beauvois and Dubois' societal concept of evaluation as "social utility". Indeed, the more power (positive SP) a person has, the more the person can be expected to be of help to the members of the group or society (s)he belongs to. There are indications that a similar societal perspective is implicitly assumed by perceivers if traits are presented in isolation, without specific referent. In that case, context-free traits have been found to be interpreted as if the referents were positive (Peeters, 1992). In this way a positive SP trait such as "strong" tends to endorse positive OP value, as in "strong friend" rather than negative OP value, as in "strong enemy". One practical consequence concerns the methodology of the test of the above hypothesis. Indeed, when comparing Obs of traits with the traits' SP and OP values, we should take care that the SP and OP measures were obtained in the same context condition as the OBs were, e.g.: both obtained when the traits were presented in isolation, as in the following experiment.
12Ninety traits from an inventory provided by Beauvois' (1996) were translated (and controlled by back translation) from French into Dutch and then rated for SP and OP by eight judges (Dutch-speaking students). Specifically, the judges rated the degree to which each trait (presented in isolation) belonged to pre-established representative standard sets of positive and negative SP and OP traits. A detailed account of this procedure, including the selection of the standard sets and their application, has been published elsewhere (Peeters, 1992). Hence we limit ourselves to a brief account that may be helpful to understand the basic principles and implications of the procedure.
13Each standard set consisted of five traits that had been found to fit alternative operational definitions of the SP and OP values of traits. For instance, the positive OP set consisted of the Dutch equivalents of tolerant, generous, sensitive, trustworthy, and trusting. In accordance with one operational definition of positive OP, those traits had been found to be selectively associated with positive (rather than negative) feelings and actions directed to others (rather than to the self). In addition, the same traits fitted a second operational definition of positive OP that was based on differential semantic interactions of OP and SP traits with the concepts "friend" and "enemy" as briefly explained yet in the theoretical introduction (Peeters 1992). In order to determine the OP value of a new trait, e.g.: intelligent, judges were asked to indicate on an 11-point scale whether the new trait belonged rather to the positive OP standard set or to an opposite negative OP standard set (intolerant, selfish, etc.). In this way each of Beauvois’ traits was situated on an SP-scale and an OP-scale, each scale ranging from –5 (most negative) over 0 to +5 (most positive). For each trait SP and OP values were averaged across judges. The reliability (Cronbach alpha) of the obtained average scores amounted to 0.73 for SP and 0.96 for OP.
14Subsequently a limited set of traits were selected to meet the following criteria at best: (1) the trait's most representative OB had to be suited for unambiguous translation in Dutch, (2) mean SP and OP values (across selected traits) had to be close to the neutral middle value (zero) of the scale, which would facilitate the interpretation of outcomes (3) standard deviations of SP and OP values across traits had to be equal, which prevented that differences obtained between correlations with SP and OP values would be caused by irrelevant differences between variabilities of SO and OP values, and (4) the correlation between SP and OP across traits had to be as low as possible making that correlations with OP and SP values would be as independent as possible. In this way, 22 traits presented in table 1 were selected. Cronbach alpha recomputed for the 22 traits were very high (0.87 for SP and 0.93 for OP).
Table 1: OP and SP values of 22 selected traits
15OBs were selected from an existing normative list of OBs associated with specific traits (Beauvois, 1996). This list was constructed in two phases. In a first phase, participants generated possible OBs, being behaviors they could enact with respect to (hypothetical possessors of) a variety of traits (e.g.: arrogant: "I try to ridicule him"). In a second phase, another group of participants selected for each trait the OB they associated most with the trait. In this way, for each trait a set of representative OBs were obtained. Within sets, OBs were ordered from most to least representative as a function of the number of participants who selected the OBs (Beauvois, 1996; personal communication).
16For each trait three OBs were selected and translated in Dutch. They represented three OB categories referred to as OB1, OB2 and OB3. OB1 was always the most representative OB according to Beauvois’ (1996) empirical criterion explained above. As the selection was done blindly, relying exclusively on the empirical data, OB1 could turn out somewhat awkward and not easy to translate in Dutch. Hence a second behaviour, OB2, was selected from the list of representative OBs taking a behaviour that did not look awkward and was most suited for translation in Dutch. OB3 was not borrowed from Beauvois’ list and did not imply concrete interaction with the target as most of Beauvois’ original OBs did. Instead it consisted simply of the act of attributing the trait to the target. It was included for the sake of ecological validity. In everyday life we often describe other people and attribute properties to them. These descriptions or attributional acts can be conceived as OBs elicited by the real or perceived properties of the described target. It seems obvious that if a target person is (perceived as) stupid, others may be inclined to describe that target as "stupid" and making that description is a genuine OB. Thus for each trait, three OBs were used. For instance, for "stingy" OB1 was "trying to change him/her", OB2 was "lending no money to him/her", and OB3 was "describing him/her as somebody who's stingy". Having 22 traits, 66 OBs were used.
17The next step was the determination of the approach-avoidance values of the OBs. For that we used an operational definition of the approach-avoidance concept established in a separate study (Peeters, 2001): Participants were asked to rate a perceiver's willingness to engage in each of 12 behaviours relative to eight hypothetical target persons of which different personality descriptions were provided. The behaviours included: wanting as friend, wanting as acquaintance, allowing to use one's belongings, helping, wanting as life-partner, collaborating with, assisting indirectly by attending others to the target's need of help, beating, scolding, ridiculizing, thwarting, and avoiding. Principal component analysis of the ratings showed that 69% of the variance was accounted for by one factor. It could be interpreted as a bipolar approach-avoidance dimension the most extreme positive and negative loadings being obtained for respectively "wanting as friend" and "avoiding".
18Using this operational definition, each OB's approach-avoidance value was determined as follows. Twelve Dutch-speaking judges (students) were presented with items presenting the OB as behaviour of a person A (the "other") directed towards B (the target). For instance "trying to change" was formulated: "A tries to change B". The judges rated the items on three 7-point rating scales indicating how high they estimated the plausibility that A would (a) want B as friend, (b) be angry with B, and (c) try to avoid B as much as possible. Notice that scales (a) and (c) are the behaviours with respectively the most extreme positive and negative loadings on the approach-avoidance factor. The scales ranged from 1 (very implausible, quasi impossible) to 7 (very plausible, quasi certain). Scale (b) inquired about possible anger directed to the target. It was added for explorative purposes because "anger" was intuitively akin but conceptually distinct from "avoidance".
19The 66 items were presented in 12 different orders in a way as to have the three OB categories (OB1, OB2 and OB3) perfectly counterbalanced and the order of items and rating scales reversed for half of the judges. Thus 12 judges provided for each OB one "approach" rating, one "avoidance" rating and an explorative "anger" rating.
20For each of the 22 traits we obtained (a) the trait's OP and SP values, and (b) the approach-, avoidance-, and anger-ratings for OB1, OB2, and OB3. For each of the ratings we computed (across N=22 traits) the partial correlations of the rating with the traits' OP value (controlling for SP) and with the traits' SP value (controlling for OP).
21The obtained partial correlations are presented in table 2. The results concerning the anger and avoidance ratings are very similar, and the approach and avoidance data are reversed mirror images of each other. This argues for the internal consistency of the data and indicates that the OBs vary along a single bipolar approach-avoidance dimension. The correlations indicate that this approach-avoidance dimension reflects the OP dimension of the traits the OBs are related to: OBs associated with positive (versus negative) OP traits have an approach (versus avoidance) character. As predicted, there seems no systematic relationship between the approach-avoidance character of Beauvois' OBs (OB1 and OB2) and the SP value of the related traits. Indeed, the partial correlations of SP with OB1 and OB2 are low, not significant, and carry inconsistent signs (-.29 vs. .09 and .19 vs. -.17). Apparently there is no relationship between the approach-avoidance character of Beauvois’ OBs (OB1 and OB2) and the SP value of the traits when possible effects of the traits' OP values are controlled for. However, otherwise than predicted, partial correlations with OB3 are significant for both OP and SP. They indicate that when a person used positive (versus negative) traits to describe a target, participants, expected the person to approach (versus avoid) the target irrespective of whether the person described the target using SP or OP traits.
Table 2 : Partial correlations of OP and SP values of traits with approach, avoidance and anger ratings of three sets of OBs associated with the traits.
22It was hypothesized that target-directed behaviours of others (OBs) that are associated with traits of the target vary along an approach-avoidance dimension. In addition, the approach (versus avoidance) value of an OB associated with a trait was expected to reflect the positive (versus negative) other profitable (OP) value of the trait rather than the trait's self-profitable (SP) value. The hypothesis was fully confirmed by the results regarding OB1 and OB2, being the concrete OBs drawn from the normative list of Beauvois (1996). These OBs are actions elicited by and directed to the target trait, e.g.: trying to change the possessor of the trait, avoiding lending money to him, etc. Quite surprisingly, a different outcome pattern was obtained for another type of OB that was referred to as OB3. An OB3 was an action for which the trait was not presented as the eliciting stimulus. Instead, the trait was presented as a substantial part of a response being the act of attributing the trait to a target. Exactly like the OBs from Beauvois' list, the act of merely attributing a trait to a target, without requiring interaction with the target, seems to have a certain approach versus avoidance value that reflects the positive versus negative OP value of the trait. However, in contrast with the approach-avoidance values of the OBs from Beauvois' list, the approach-avoidance value of the trait attribution reflects the positive versus negative SP value of the trait as well. What this means is illustrated by the following example.
23Let us consider the traits "ambitious" and "anguished". As it is shown in Table 1, they carry nearly similar OP values but opposite SP values, ambitious being positive and anguished being negative. Because of the equivalence of the traits' OP values they should elicit OBs with nearly similar approach-avoidance values. This seemed effectively the case for the concrete OBs (OB1 and OB2) drawn from the normative list of Beauvois, which included, for instance, discussing with (the ambitious target) and trying to reassure (the anguished target). Judges assigned nearly the same low avoidance values to both OBs (the rounded averages amounted to 2). However, the same judges assigned different approach-avoidance values to the OBs of type OB3 that consisted of attributing the traits. In this way the chance that A would avoid B was rated rather high (rounded average: 5) if A described B as "anguished", while the chance was rated rather low (rounded average: 3) if A described B as "ambitious".
24Apparently, when the OP value of the traits was controlled, subjects assumed that a perceiver who uses positive SP traits to describe a target expresses a higher approach (lower avoidance) tendency relative to the target than a perceiver who uses negative SP traits. This observation may have important theoretical consequences. It suggests that a distinction should be made between evaluative meaning as a perceived incentive value stimulating approach-avoidance responses on the one hand, and evaluative meaning as a perceived expressive value revealing approach-avoidance response sets on the other. Specifically the perceived incentive value of a trait’s evaluative meaning resides in the fact that the trait is expected to elicit a correspondent approach-avoidance tendency directed to the possessor of the trait. The expressive value resides in the fact that someone who uses the trait to describe a target is expected to harbour correspondent approach-avoidance tendencies directed to the target.
25The present results suggest that evaluative meaning as incentive value for approach-avoidance is processed in accordance with the well-established two-dimensional model according to which a likeability-related evaluative dimension (OP) should be distinguished from a power-related evaluative dimension (SP). Indeed, the approach-avoidance incentive value seems limited to the likeability-related evaluative meaning. However, evaluative meaning as expressive value seems processed in accordance with a simple unidimensional evaluative meaning model within which the distinction between likeability-related OP and power-related SP has been reduced both appearing as functionally equivalent implementations of the same good-bad dimension. Indeed, irrespective of whether traits are likeability- or power-related, positive traits express approach and negative traits express avoidance.
26The present difference between traits’ incentive and expressive values for approach-avoidance was unexpected. A possible explanation may be looked for in theory and research on actor-observer effects (Jones & Nisbett, 1972). The OBs of Beauvois and Dubois (OB1 and OB2) were generated by students who took the perspective of an actor involved with the target. In comparison with observers, actors can be expected to be more focussed on situational determinants of the action such as traits of the target. This may result in the formation of more complex representations of the target involving two, rather than one, evaluative trait dimension, and this complex representation may transpire into the OBs generated from the actor's perspective. However, the OB3 may not reflect that more complex representation because it was not generated from the perspective of an actor but simply constructed by the experimenter as a description a possible actor made of the target. The subjects who rated the approach-avoidance tendency that was expressed by the actor's description of the target, may have processed the traits following the simple unidimensional model because they took the observer's perspective.
27Since Asch’s (1946) seminal study it has been evidenced that impressions of personality are composed of traits that are organized consistent with relatively simple structural cognitive models. For instance, Rosenberg and Sedlak (1972) advanced a two-dimensional model involving two evaluative meaning dimensions that could be matched with the power-related "SP" and likeability-related "OP" dimensions explained in the introductory section and used in the subsequent empirical study. Traditionally traits have been conceived as behavioural dispositions of the possessor of the traits. Beauvois and Dubois (2000; Dubois and Beauvois, 2001), however, demonstrated that the evaluative meaning of a trait relates to the trait’s affordances, dealt with as behavioural dispositions of others relative to the possessor of the trait or "OBs". In the present study it has been shown that, analogously to the cognitive organization of the traits, also the cognitive organization of the OBs is underlain by a simple structure being a dimension contrasting a tendency to approach against a tendency to avoid the possessors of the traits. Moreover, the simple structure underlying OB organization can be matched with part of the simple structure underlying trait organization, at least as far as the traits function as incentives for OB. Indeed, the approach-avoidance "OB" response reflects the evaluative OP dimension and is unrelated to the SP dimension: people with positive OP traits such as "sociable" are approached and wanted as friends, while people with negative OP traits such as "violent" are avoided. At the same time, people with positive SP traits such as "ambitious" are neither more approached nor more avoided than people with negative SP traits such as "anguished".
28However, traits do not exclusively function as incentives. If traits are used by a communicator in order to describe a target, then the evaluative meaning of the traits expresses the communicator’s attitude towards the target. The present data suggest unexpectedly that the approach-avoidance tendencies associated with those attitudes reflect both the OP and the SP value of the traits. This means that a communicator may express a positive attitude and willingness to approach not only by describing a target as sociable (positive OP), but also by describing the target as ambitious (positive SP). However an audience that hears the description may respond to the incentive value of the traits and feel stimulated to approach the target only if the target is described as sociable, and not if it is described as ambitious. If future research confirms the present findings, we may not only have traced a source of potential misunderstanding in human communication. In addition, we may have advanced our knowledge about possible conditions determining whether the cognitive processing of traits involves two distinct evaluative dimensions or one single encompassing good-bad dimension.