29.01.2013 Views

S1 (FriAM 1-65) - The Psychonomic Society

S1 (FriAM 1-65) - The Psychonomic Society

S1 (FriAM 1-65) - The Psychonomic Society

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Posters 2055–2061 Friday Noon<br />

problem, we present a range of reinforcement learning models that explore<br />

how slight shifts in attended dimensions and reinforced units of<br />

behavior can result in different allocation strategies. Beyond mimicking<br />

aggregate human data, our models explain how sequential<br />

decision making can be understood as a combination of feedback- and<br />

strategy-driven processes, effectively reconceptualizing probability<br />

matching as representational learning.<br />

(2055)<br />

Exploring the Supraliminal Mere Exposure Effect. MARC G.<br />

BERMAN, RICHARD GONZALEZ, & MICHAEL S. FRANKLIN,<br />

University of Michigan—<strong>The</strong> mere exposure effect is a phenomenon<br />

where participants prefer stimuli that have been repeatedly presented<br />

as opposed to stimuli that have not been presented. <strong>The</strong> stimuli in most<br />

mere exposure studies are neutral and novel to the participant, therefore<br />

eliminating preexisting valence. Here, we explored how the<br />

supraliminal mere exposure effect was affected by differently valenced<br />

stimuli (as determined by independent raters). In addition, we<br />

explored potential factors accounting for the observed heterogeneity<br />

whereby some participants show the mere exposure effect and others<br />

do not. Furthermore, we present a statistical method for analyzing<br />

mere exposure data using ordinal regression that is more sensitive<br />

than previously utilized techniques and permits modeling the heterogeneity.<br />

Preliminary findings suggest that subjects who do show the<br />

effect are not influenced by stimulus valence and utilize recognition<br />

to determine stimulus preference. <strong>The</strong>refore, it appears that recognition<br />

overrides initial stimulus valence in supraliminal mere exposure<br />

preference judgments.<br />

(2056)<br />

A Comparison of Reaction-Time Models of Multiple-Choice Decisions.<br />

FABIO LEITE & ROGER RATCLIFF, Ohio State University<br />

(sponsored by Roger Ratcliff)—Several sequential sampling models<br />

for multiple-choice decisions were evaluated. <strong>The</strong>se models used various<br />

racing diffusion processes (e.g., Usher & McClelland’s, 2001,<br />

leaky competing accumulator model). <strong>The</strong> structures of the models<br />

differed on a number of dimensions, including starting point of competing<br />

accumulators, lateral inhibition, and constraints on accumulation<br />

rates (e.g., accumulators only allowed to take nonnegative values).<br />

Data were collected from a letter-discrimination experiment in<br />

which stimulus difficulty or probability of the response alternatives<br />

was varied along with number of response alternatives. Performance<br />

of the models was addressed by testing them against empirical data,<br />

including issues of psychological plausibility of parameter estimates<br />

and mimicry among models.<br />

(2057)<br />

Effects of Caffeine on Deck Selection in Two Gambling Tasks.<br />

PATRICK A. RAMIREZ & DANIEL S. LEVINE, University of Texas,<br />

Arlington—Participants were given two different gambling tasks: the<br />

Iowa Gambling Task (IGT) due to Bechara et al., and a variant of the<br />

modified IGT due to Peters and Slovic, in either order, with 100 trials<br />

in each. On the second task, the participants were given either a caffeinated<br />

drink (treatment), a decaffeinated drink (placebo), or water<br />

(control). It was found that caffeine significantly increased the tendency<br />

to choose advantageous decks, but only on the Bechara task and<br />

not on the Peters task. <strong>The</strong> advantage of caffeine on the Bechara task<br />

was manifest even on the first 20 deck selections. This led us to conjecture<br />

that the effect of caffeine could be primarily an enhancement<br />

of attention, with the Bechara task being the more subject to inattention<br />

because the Peters task has more variability in its outcomes.<br />

(2058)<br />

Human and Optimal Valuation in a Sequential Decision-Making<br />

Task With Uncertainty. KYLER M. EASTMAN, BRIAN J. STANKIE-<br />

WICZ, & ALEX C. HUK, University of Texas, Austin—Many sequential<br />

sampling models suggest that decisions rely on the accumulation<br />

of evidence over time until reaching a particular threshold.<br />

77<br />

<strong>The</strong>se models can often account for variations of speed and accuracy<br />

in perceptual tasks by manipulation of this threshold. But how does<br />

this threshold get determined in real-world decisions? It has been hypothesized<br />

that the threshold maximizes some reward function, possibly<br />

incorporating measurements of both speed and accuracy (Gold<br />

& Shadlen, 2003). This approach has produced a family of models that<br />

accurately describes behavior for two-alternative forced choice<br />

(2AFC) tasks. (Bogacz et al., 2006) However, it has been unclear what<br />

the optimal threshold becomes when additional perceptual information<br />

can be obtained at a cost. We present a model of optimal sequential<br />

decision making in a task that extends the traditional 2AFC<br />

by adding the option of additional information. In the task, the observer<br />

receives a sample from two overlapping distributions.<br />

• JUDGMENT •<br />

(2059)<br />

Associations Between Absent Events in Contingency Judgment.<br />

LEYRE CASTRO, FABIÁN A. SOTO, & EDWARD A. WASSERMAN,<br />

University of Iowa—Studies of contingency judgment have shown that<br />

cue–outcome associations can be strengthened or weakened even<br />

when the cues are absent (e.g., retrospective revaluation effects). But,<br />

there is no evidence in the mainstream contingency judgment literature<br />

for the formation of new associations between an absent cue and<br />

an absent outcome. Interestingly, research with human infants<br />

(Cuevas, Rovee-Collier, & Learmonth, 2006) and nonhuman animals<br />

(Dwyer, Mackintosh, & Boakes, 1998) has disclosed that new associations<br />

between cues and outcomes can be formed when neither of<br />

them is presented. In our own contingency judgment experiments, we<br />

have found evidence for the formation of new associations between<br />

cues and outcomes that are absent, but that are associatively activated.<br />

Our results raise difficult problems for cognitive or inferential accounts<br />

(e.g., De Houwer, Beckers, & Vandorpe, 2005), but they can<br />

be readily explained by associative learning models (e.g., Dickinson<br />

& Burke, 1996; Van Hamme & Wasserman, 1994).<br />

(2060)<br />

<strong>The</strong> Effect of Race on Guilty Verdicts: Order Matters. LESLIE N.<br />

KNUYCKY, HEATHER M. KLEIDER, & ASHLEY MYERS, Georgia<br />

State University (sponsored by Stephen D. Goldinger)—Factors<br />

other than case evidence (i.e., available cognitive capacity, judgment<br />

instructions) influence juror verdicts (Goldinger et al., 2003). Racial<br />

bias, one factor shown to affect jury judgments (Jones & Kaplan,<br />

2003), is posited as an automatic response often difficult to intentionally<br />

control. We examined whether attempts to inhibit bias may<br />

“backfire” when race is made especially salient. In two experiments,<br />

case presentation order (varying position of black or white defendants)<br />

and instructions to ignore race impacted jury decisions. <strong>The</strong> results<br />

showed that with no instruction, black defendants received<br />

harsher judgments when presenting white- before black-defendant<br />

cases. This order effect was negated in Experiment 2 with instruction<br />

to disregard race, as guilty verdicts increased overall for black defendants.<br />

This suggests that when race is “noticeable” via presentation<br />

order or instruction, racially biased decisions result. Such decisions<br />

lead to more severe verdicts for black defendants than when race goes<br />

“undetected.”<br />

(2061)<br />

Judgments of Randomness: Is <strong>The</strong>re a Bias Toward High Alternation?<br />

FIONA E. J. MCDONALD & BEN R. NEWELL, University of<br />

New South Wales (sponsored by Ben R. Newell)—Extensive research<br />

has documented that people are poor generators of random sequences.<br />

A key feature of human generated sequences is a tendency for excessive<br />

alternation of outcomes. <strong>The</strong> research on judgments of randomness<br />

is not as extensive but it has been found that when judging sequences,<br />

there is also a tendency to judge those with excessive<br />

alternation as most random. We challenge this notion, finding through<br />

a series of experiments only minimal bias in judgments of random-

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!