Abstracts 2005 - The Psychonomic Society

Posters 3098–3104 Friday Evening
also extend recent work on the benefits of multimedia learning (Gyselinck
et al., 2002; Mayer, 2001).
(3098)
The Role of Conceptual Relations in Recognition Memory. LARA L.
JONES, ZACHARY ESTES, & RICHARD L. MARSH, University of
Georgia—We assessed whether conceptual relations (WOOL BLANKET:
made of ) facilitate recognition memory for the constituent concepts
(WOOL and BLANKET), as suggested by the relational information hypothesis
(Humphreys, 1976). In each experiment, participants judged
the sensicality of word pairs prior to a surprise recognition memory test.
Experiment 1 showed that recognition memory for word pairs was facilitated
by relational information. In Experiment 2, relational information
facilitated memory for the modifier (WOOL), but not for the head
noun (BLANKET), and only when the same relation was instantiated at
study and test. Experiment 3A replicated this result with individual,
rather than paired, presentation of words at test. When the memory test
was not a surprise, in Experiment 3B, conceptual relations did not facilitate
memory. Results support the relational information hypothesis
and suggest that relational information is more strongly associated with
the modifier than with the head noun (Gagne & Shoben, 1997).
• REASONING •
(3099)
The Role of Interpretation in Causal Learning. CHRISTIAN C.
LUHMANN, Vanderbilt University, & WOO-KYOUNG AHN, Yale
University (sponsored by Woo-Kyoung Ahn)—When learning causal
relations from covariation information, do people give causal interpretations
to covariation as it is encountered, or is covariation simply
recorded verbatim for later causal judgments? Participants in Study 1
performed a causal learning task and were asked to periodically interpret
observations. Participants’ interpretations were context sensitive;
identical observations were interpreted differently when encountered
in different contexts. These interpretations also predicted
their final causal strength judgments. Those who were influenced
more by initial covariation (primacy effect) tended to interpret later
observations as consistent with initial covariation. Conversely, those
who were influenced more by later covariation (recency effect) tended
to interpret later observations as consistent with later covariation.
When participants in Study 2 were required to predict the presence of
the effect on each trial, overall recency effects were obtained, but interpretations
were still context sensitive and predicted causal strength
judgments. We discuss the difficulty these results pose for current
models.
(3100)
The Influence of Virtual Sample Size on Confidence in Causal
Judgments. MIMI LILJEHOLM & PATRICIA CHENG, UCLA—We
investigated the extent to which confidence in causal judgments varied
with virtual sample size—the frequency of cases in which the outcome
is (1) absent before the introduction of a generative cause or
(2) present before the introduction of a preventive cause. Participants
were asked to evaluate the influence of various candidate causes on
an outcome, as well as their confidence in judgments about those influences.
They were presented with information on the relative frequencies
of the outcome, given the presence and absence of various
candidate causes. The study manipulated these relative frequencies,
sample size, and the direction of the causal influence (generative vs.
preventive). We found that confidence varied with virtual sample size
and that both confidence and causal strength ratings showed a clear,
and previously undocumented, asymmetry between the two causal
directions.
(3101)
The Useful Application of a Discounting Heuristic in Causal Inference.
KELLY M. GOEDERT, Seton Hall University, & SARAH A.
LARSON, Pacific Lutheran University—Previously, we demonstrated
102
that people reduce their causal effectiveness ratings of a moderately
effective cause in the presence of a highly effective one (i.e., discounting;
Goedert & Spellman, 2005). Here, we assess whether participants
recruit this discounting heuristic because it is useful. In positive
outcome situations, finding one cause may be good enough. With
negative outcomes, one must find all possible causes to eliminate an
aversive event. Participants received contingency information about
potential causes of either a positive outcome (medications leading to
allergy relief) or a negative outcome (environmental irritants causing
allergies) over 72 trials. In both situations, a moderately effective
cause was learned about in the presence of an alternative that was either
highly effective or not at all effective. Participants discounted the
moderately effective cause in positive, but not negative, outcome situations.
These results suggest that participants selectively recruit the
discounting heuristic when finding that only one cause is sufficient.
(3102)
Darwin and Design: Do Evolutionary Explanations Require Understanding
Evolution? TANIA LOMBROZO, Harvard University (sponsored
by Yuhong Jiang)—Three experiments examined the relationship
between knowledge of causal mechanisms and explanatory understanding.
In Experiment 1, 64 participants identified statements that
would falsify a functional explanation for an artifact (a product of
human design) or a biological part (a product of natural selection). Although
participants judged functional explanations for artifacts and
biological parts equally satisfying, they were significantly worse at
identifying statements that falsify evolutionary explanations. Experiments
2 (N = 72) and 3 (N = 36) showed that participants’ judgments
of the goodness of evolutionary explanations are uncorrelated with
their ability to identify statements that falsify the explanation and that
the inability to identify falsifying statements results from a poor understanding
of the mechanisms by which natural selection operates.
The feeling of understanding that accompanies an evolutionary explanation
may result from knowledge that an underlying mechanism
that parallels human design exists (namely, natural selection), no matter
that the mechanism is not understood.
(3103)
Priming Category and Analogy Relations During Relational Reasoning.
ADAM E. GREEN, JONATHAN A. FUGELSANG, KEVIN A.
SMITH, & KEVIN N. DUNBAR, Dartmouth College—Green, Fugelsang,
and Dunbar (2004) have demonstrated that the processing of
analogies primes two kinds of underlying relations, category and analogy
relations, and that these primed relations have an effect on subsequent
tasks. In Green et al. (2004), we found that this priming interfered
with performance in a Stroop task. Here, using a naming task and the
same stimuli, we hypothesized that processing of analogies should facilitate
reading of words that refer to category and analogy relations.
Results were consistent with this hypothesis, and we discuss the results
with respect to models of analogical reasoning and priming effects.
(3104)
Cue Competition With Continuous Cues and Outcomes. JASON M.
TANGEN, BEN R. NEWELL, & FRED WESTBROOK, University of
New South Wales (sponsored by Ben R. Newell)—The simplest measure
of causal inference is based on the presence or absence of causes
and effects. Traditionally, animal and human contingency learning experiments
have investigated the role of these binary events that do not
capture the nature of most everyday assessment tasks. In a basic twophase
design (Phase 1, A�/B�; Phase 2, AC�/BD�), animals tend
to respond less to C (blocking) than to D (superconditioning). We employ
this design with human participants and use continuous cues and
outcomes, rather than binary events (i.e., different quantities of liquid
resulting in various levels of plant growth). Under these conditions,
we demonstrate that the differential treatment of C and D in the second
phase disappears. Although conditional contingency has been discussed
at length with respect to binary events, we will discuss the role
of its continuous analogue in human covariation assessment.