S1 (FriAM 1-65) - The Psychonomic Society

Friday Morning Papers 8–14
focal attention is allocated to the feature singleton. These results are inconsistent
with theories that claim that it is possible to detect a feature
singleton without directing attention to the location of the singleton.
9:00–9:15 (8)
Numerical Format Effects in the Eriksen Task. PHILIP T. QUINLAN,
University of York—Although this is not the only study to have examined
numerical versions of the Eriksen flanker task, it is (probably)
the only study with displays containing both Arabic and Chinese characters.
Twenty-eight Chinese/English bilingual undergraduates were
tested in two numerical versions of the task. On every trial, subjects
were timed to classify a central target—either four/six or five/seven—
in a row of three characters. Across the versions the target was either
always Arabic or Chinese. Flankers were either both Arabic, or both
Chinese. On Same trials all the characters represented the same number,
on Congruent trials the target and flankers differed but each was
assigned to the same response, on Incongruent trials the targets and
flankers demanded different responses. Robust slowing on incongruous
trials was found throughout. The effect was enhanced when the
targets were Chinese. Both notions of unattended processing and theories
of numerical cognition are considered.
9:20–9:35 (9)
A General Computational Theory of the Distribution of Visual
Spatial Attention. GEORGE SPERLING, IAN J. SCOFIELD, &
ARVIN T. HSU, University of California, Irvine—We derive a computational
theory of the distribution of visual attention using a linear
systems approach. First, we measure an observer’s ability to distribute
attention sinusoidally along rows or columns in a 12 � 12 array
that contains 1 target (a large disk) on 1 of 72 attended locations, 10
false targets among 72 unattended locations (to force the observer to
ignore unattended locations), and distractors (small disks) elsewhere
(Gobell, Tseng, & Sperling, 2004). These data then enable the theory
to make accurate, completely parameter-free predictions of the same
observer’s ability to distribute spatial attention in response to arbitrarily
complex 72-square requested patterns of attentional distribution.
The theory contains (1) a spatial acuity function, (2) an attention
modulation-transfer function that describes the decline of attentional
conformability with increasing spatial frequency, (3) multiplicative
combination of (1) and (2), (4) random decision noise, and (5) a decision
process that selects the most likely target location.
9:40–9:55 (10)
Using Foils to Measure Spatial Tuning Functions for Visual Attention.
JOHN PALMER, University of Washington, & CATHLEEN
M. MOORE, University of Iowa—Our goal is to measure the spatial
extent of visual attention. To do so, observers are required to detect a
visual target while ignoring a nearby foil that is identical to the target
except for location. The experiment includes a manipulation of both
the contrast of the foil and the separation between the foil and the relevant
location. The appropriate measure of selectivity depends on how
attention modulates the effect of contrast. Two common hypotheses
for attentional modulation are contrast gain versus an all-or-none mixture.
The results disconfirm the contrast gain hypothesis and are instead
consistent with the all-or-none hypothesis. Moreover, the effect
of the separation between the foil and the relevant location is very
large: Performance ranges from chance to perfect. We are now using
the foil task to measure spatial tuning functions for visual attention.
Models of Choice and Decision Making
Beacon A, Friday Morning, 8:00–10:00
Chaired by Jerome R. Busemeyer, Indiana University
8:00–8:15 (11)
A Quantum Information Processing Explanation of Disjunction
Effects. JEROME R. BUSEMEYER, Indiana University, ZHENG
WANG, Ohio State University, & MERV MATTHEW, Indiana Uni-
2
versity—A new approach to decision theory based on quantum information
processing principles is used to explain some paradoxical phenomena
of human choice behavior. Quantum strategies were originally
used to explain the fact that humans prefer to cooperate rather
than defect in a Prisoner Dilemma game. Here, we develop a quantum
model for the disjunction effect. This refers to a paradox in which (1) a
player prefers to defect when the player knows that an opponent will
defect, and (2) the player also prefers to defect when the player knows
that an opponent will cooperate, but (3) the player reverses preference
and cooperates when the opponent’s action is unknown. New experimental
findings on the disjunction effect are reported, and a quantum
explanation for the findings is presented. The quantum model is also
compared to traditional information processing models.
8:20–8:35 (12)
The Simplest Model of Choice and Reaction Time. SCOTT D.
BROWN & ANDREW J. HEATHCOTE, University of Newcastle—
Over the past 50 years, accumulator models of response time and accuracy
have become increasingly sophisticated, to accommodate an increasing
range of empirical phenomena. In the last 2–3 years, an effort
has been made to identify models that either (1) are just as powerful as
their more complex competitors, but a little simpler or (2) have less explanatory
power, but are much simpler. Brown and Heathcote (2005,
Psychological Review) developed a ballistic accumulator model, of
Type A. We now proceed one step further, and demonstrate that an even
simpler model can accommodate almost all the major benchmark phenomena,
and provides a very good fit to real data. This linear BA
model has the unique advantage among its competitors of completely
analytic solutions for it’s predicted RT distributions, even for choices
between many alternatives (2+).
8:40–8:55 (13)
Gaining Insights From Reaction Time Data Using Bayesian
Wiener Diffusion. JOACHIM VANDEKERCKHOVE & FRANCIS
TUERLINCKX, University of Leuven, MICHAEL D. LEE, University
of California, & ERIC-JAN WAGENMAKERS & GILLES DUTILH,
University of Amsterdam (read by Francis Tuerlinckx)—The Wiener
diffusion process is a convenient model for the analysis of two-choice
response time data. We use Bayesian methods to extend the applicability
of the diffusion model. The extended methods are easy to apply,
show no serious numerical problems, and yield output that can be interpreted
in a direct and intuitive way. The Bayesian context also allows
for the easy evaluation of hypotheses that would otherwise be
very complex to test. Combined with the straightforward interpretation
of the diffusion model’s parameters, this type of analysis will enable
researchers to investigate cognitive processes in ways that were
not previously practicable. Fitting functional forms to parameters,
modeling changes in parameters over time, and capturing random effects
across individuals are just some examples. We have successfully
implemented the Wiener diffusion model in WinBUGS, and will illustrate
the potential of the method with an experiment investigating
practice effects in a lexical decision task.
9:00–9:15 (14)
Dual Processes and Development: Explaining Contradictory Relations
Between Risk Perception and Risk Taking. VALERIE F.
REYNA, BRITAIN MILLS, STEVEN ESTRADA, & CHARLES J.
BRAINERD, Cornell University—Dual processes in fuzzy-trace theory
predict opposite relations between risk perception and risk taking:
Both positive and negative correlations were obtained within the same
individuals, depending on verbatim versus gist cues in questions.
These results might mask developmental changes in such effects,
however. We tested this hypothesis by assessing relations between risk
perception and risk taking separately for different ages in a large sample
(N = 596) of 14- to 17-year-olds. Measures of risk perception differed
in cue specificity and response format. Measures that emphasized
verbatim retrieval and quantitative processing produced positive
correlations, but these relations did not change monotonically with