- In this study we examined how certain structural properties of social
explanations contribute to their perceived coherence. Specifically, we investigated the
possible influence of "causal syntax" on perceptions of explanation quality.
According to the causal syntax notion, a coherent causal scenario consists of different
types of causal elements connected together according to implicit rules that govern how
elements can be causally linked (Read & Marcus- Newhall, 1993; Read, 1987; 1992).
Explanations are perceived as less coherent if these rules are contradicted or are
incompletely applied.
- Violations of causal syntax can occur in everyday discourse for a
variety of reasons. For example, we sometimes leave out intervening steps in communicating
an explanation if we believe they might be easily inferred by the listener (Read, 1987;
1992). In explaining why a store manager had a vandal arrested, we might say it was
because the vandal broke the store window by throwing a rock. Although we might not also
say that the store manager was angry, we assume the listener will infer that breaking the
window initiated the mental state of anger in the store manager and his anger was a reason
for having the vandal arrested. To the extent that intervening inferences are ambiguous or
otherwise difficult, the listener may regard an explanation as less coherent than the
explainer intended.
We reasoned that if given the opportunity to
improve an explanation, perceivers would add information according to the causal syntax
formulation. For instance, one way to strengthen an explanation would be to specify the
link between the initiating act and the mental state that led to the actor's behavior.
Another might be to clarify how a particular mental state can be a reason for the actor's
behavior, such as how anger leads to retaliatory behavior or punishment.
METHOD
We presented 24 participants (12 men and 12 women)
with 12 causal explanations for each of three behavioral events and instructed them to
modify an explanation if they felt they could make it a better account for the event. The
behavioral events and causal explanations were the same as those used in previous studies
(Sherman et al., 1995). Based on previous norm data, half the explanations for each event
were normatively high on dimensions of plausibility, sufficiency, and likelihood, and the
remainder moderate. Within each level of quality, half the explanations implied an actor
causal locus and half implied an entity causal locus.
RESULTS
- We first categorized each modification according to whether it
focused on the actor (A) or on the entity (E) involved in the event. Next, we coded the
actor-focus modifications in terms of whether they involved explications of the link
between initiating conditions and actors' internal states (S) or between internal states
and the actors' actions (A). We categorized the entity-focused modifications as attempts
to clarify either the initiating power of the entity to produce an internal state in the
actor (S), or the power of the entity to produce the action directly in the actor (A). The
frequencies of each type of modification are shown in Figure 1.
- Actor/State modifications were clearly the most frequent type in all four locus/quality
categories of explanations, with the most striking preference occurring for moderate
quality entity explanations. Further, the Actor/State modifications were more frequently
made to moderate quality than high quality accounts. This quality effect, which was
significant for both actor oriented reasons, Chi Square (1, N=75) = 7.05, p < .01 and
for entity oriented reasons, Chi Square (1, N=106)= 18.26, p < .01, was particularly
pronounced in the case of entity-oriented explanations (cf. Figure 1).
-
For the remaining types of modifications there was a
general tendency for participants to make more modifications to moderate than high quality
explanations, and to do so with a modification in which the focus (Actor or Entity)
matched the causal locus implied by the explanations (also Actor or Entity). For example,
Actor/Action modifications were significantly more frequent for moderate than high quality
actor oriented accounts, Chi Square (1, N=43) = 10.44, p < .01, but did not differ for
high and moderate entity accounts, Chi Square (1, N=32) = 2.0, n.s. This pattern was
reversed for Entity/State and Entity/Action repair types.
DISCUSSION AND IMPLICATIONS
Participants' predominant strategy for improving
moderate quality explanations was to add information that identified an actor's internal
state as a plausible reason for the event, and this strategy was particularly likely for
explanations in which the causal locus was implied to be something external to the actor.
It can be noted that this strategy fills in missing or ambiguous links in a causal chain
that begins with an initiating or enabling state and ends with the actor's observed
behavior, and thus is quite consistent with the general notion of causal syntax. A
secondary strategy, which seems more akin to bolstering or clarifying links, was for
participants to provide information that reinforced the plausibility of the implied causal
locus, for example by adding information regarding an entity's power to produce an
internal state in the actor or an entity's power to influence the actor's behavior
directly.
REFERENCES
- Read, S.J. (1987). Constructing causal scenarios: A knowledge
structure approach to causal reasoning. Journal of Personality and Social
Psychology, 52, 288-302.
- Read, S.J. (1992). Constructing accounts: The role of explanatory
coherence. In McLaughlin, M.L., Cody, M.J., & Read, S.J. (Eds.), Explaining
Oneself to Others: Reason-Giving in a Social Context (pp. 2-20). Hillsdale, N.J.:
Erlbaum.
- Read, S. J., & Marcus-Newhall, A. R. (1993). Explanatory
coherence in social explanations: A parallel distributed processing account. Journal
of Personality and Social Psychology, 65, 429-447.
- Sherman, Richard C., Lim, K.M., Seidel, S.D., Sinai, K.A., &
Newman, K.M. (1995). Processing causally relevant information. Journal of
Personality and Social Psychology, 68, 365-376.
-
Request Reprint (Include Your Surface Mail
Address)
Return to First Page