Abstracts of the Psychonomic Society — Volume 11 — November 2006 47th Annual Meeting — November 16–19, 2006 — Houston, Texas Papers 1–8 Friday Morning Semantic Priming within-language associative priming effects were quite similar. Fi- Grand Ballroom JKL, Friday Morning, 8:00–9:20 nally, Experiment 3 showed a significant translation priming effect for noncognates. Thus, highly fluent bilinguals do develop between- Chaired by David A. Balota, Washington University language links with noncognates at the conceptual level. We examine the implications of these results for models of bilingual memory. 8:00–8:15 (1) Effects of Semantic Priming and Stimulus Quality: Insights From Judgment and Decision Making 1 RT Distributional Analyses. DAVID A. BALOTA & MELVIN J. YAP, Grand Ballroom ABC, Friday Morning, 8:00–10:00 Washington University—In a series of experiments, reaction time (RT) distribution analyses were used to examine the nature of the in- Chaired by Christine Ruva, University of South Florida fluence of semantic priming on lexical decision and word-naming per- formance. In contrast to variables that both shift and skew the RT dis- 8:00–8:15 (5) tributions (e.g., word frequency and lexicality), semantic relatedness Judgments of Functional Relationships in Systems of Three Vari- produces primarily a shift in the RT distribution, with no increase in ables. KENT L. NORMAN & BENJAMIN K. SMITH, University of skewing. In the standard clear target conditions, the shifting of the RT Maryland, College Park—The participants were told the relationship distribution suggests a constant influence of priming on lexical pro- between variables A and B and between B and C and then asked to cessing across all targets. In contrast to this shifting, the interactive judge the relationship between A and C. The order, the names, and the effects of target degradation and semantic priming primarily occur at relationships (increasing, decreasing, and no change) were varied sys- the tail of the RT distribution, suggesting an additional process in- tematically and resulted in 72 judgment problems. Four groups of 15 voked with degraded targets. Interestingly, this pattern was also found participants each made judgments for fictitious systems of variables when primes were highly masked. The discussion focuses on single in chemistry, psychology, or economics. An additional chemistry versus multiple process models of semantic priming. group was allowed the option “don’t know.” Although there were no mathematically correct answers, the results showed consistent pat- 8:20–8:35 (2) terns of inference (transitivity, bidirectionality, and solvability), but On a PDP Model of Lexical Processing: More Words About Stages. no effect due to system. For example, positive relationships between DEREK BESNER, SZYMON WARTAK, & SERJE ROBIDOUX, A and B and between B and C resulted in a positive relationship be- University of Waterloo—Plaut and Booth (2006 in Psych. Rev. and tween A and C (transitivity and bidirectionality). A number of other Plaut & Booth 2000, also in Psych. Rev.) assert that their PDP model consistent inferences were found for mixed relationships that are of lexical processing simulates the repeated observation, in the con- being used to formulate a theory of inference for systems of variables. text of lexical decision, that (1) stimulus quality and semantic relat- edness interact on RT, but that (2) stimulus quality and word fre- 8:20–8:35 (6) quency have additive effects on RT. The results of new simulations The System of Concepts Behind the Simplest Medical Diagnosis. with their model undermine this conclusion. ROBERT M. HAMM, University of Oklahoma—The simplest situa- tion for medical diagnosis and treatment involves one disease, one di- 8:40–8:55 (3) chotomous test for it, and one treatment for it. However, seven pri- Relational Integration in Lexical Processing. ZACHARY ESTES, mary concepts and eight derived concepts are applicable to this University of Warwick, & LARA L. JONES, University of Georgia— situation. Psychologists have studied physicians’ judgments of most Relational integration is the inference of some semantic relation (e.g., of the primary concepts (pretest probability, sensitivity, specificity, containment) between two concepts (e.g., BEAN and JAR), such that utility of correct, missed, and unnecessary treatments and of correctly they jointly refer to a single object or event (BEAN JAR). We investi- not treating), but only a few of the derived concepts (probabilistic in- gated the influence of relational integration on lexical processing. In ference: posttest probability via Bayes’s Theorem; treatment deci- Experiment 1, lexical decisions for the exact same word pairs were sions). Judgments of other concepts (e.g., utility impacts of unneces- faster when they were easily integrated (e.g., HORSE DOCTOR) than sary treatment and of missed treatment; action threshold, no-test/test when they were more difficult to integrate (e.g., DOCTOR HORSE). In threshold, and test/treat threshold probabilities; choice whether to Experiment 2, this “integrative priming” was obtained across varying test, treat, or neither) have been little studied. How accurate are they? proportions of related prime–target trials. Experiments 3 and 4 used By what processes are they made? How may they be aided? Consider- a variable prime–target SOA to compare the temporal dynamics of in- ation of the full system of concepts provides a context in which the prac- tegrative priming with those of associative and semantic priming. In- tical importance of improving probabilistic inference can be realized. tegrative priming was dissociated from associative priming at a 2,000-msec SOA (Experiment 3), but closely resembled semantic 8:40–8:55 (7) priming from early (100-msec) to late (2,500-msec) SOA conditions. Testing a Class of Utility Models. RICHARD A. CHECHILE & Relational integration appears to act prospectively and occur inevitably. SUSAN F. BUTLER, Tufts University—The Miyamato (1988) generic utility theory (GUT) is a utility representation that captures a wide 9:00–9:15 (4) class of other utility models. Moreover, the entire class of models can Masked Associative/Semantic and Identity Priming Across Lan- be evaluated by examining a key parameter that emerges whenever guages With Highly Proficient Bilinguals. MANUEL PEREA, Uni- there are mixed gambles, i.e., a gamble with a possible gain and a pos- versitat de València, & JON A. DUÑABEITIA & MANUEL CAR- sible loss. In a comprehensive experiment, it is shown that this para- REIRAS, Universidad de La Laguna—One key issue for models of meter varies in a systematic fashion that is inconsistent with the GUT bilingual memory is to what degree the semantic representation from class of models. The implications of the experiment are developed for one of the languages is shared with the other language. Prior research a theory of risky choice. has suggested that there are shared conceptual representations for cog- nates, but not for noncognates. However, there is no reason a priori to 9:00–9:15 (8) consider that the conceptual representation of noncognates differs Response Time Tests of Take-the-Best and Rational Models of De- from that of cognates. Experiment 1 was a between-language masked cision Making. ROBERT M. NOSOFSKY & F. BRYAN BERGERT, associative priming lexical decision experiment in which we used Indiana University, Bloomington—We develop and test generalized noncognate pairs with highly fluent bilinguals. The results showed a versions of take-the-best (TTB) and rational (RAT) models of multi- significant between-language associative priming effect. Experi- attribute paired-comparison inference. The generalized models make ment 2 showed that the magnitudes of the between-language and allowance for subjective attribute weighting, probabilistic orders of at- 1 Friday Morning Papers 9–14 tribute inspection, and noisy decision making. The key new test in- were studied on blue and red backgrounds. Conjunction test items were volves a response time approach. TTB predicts that response time is constructed from parts of faces that had been either the same or dif- determined solely by the expected time required to locate the first dis- ferent colors during the study phase. The subjects made more false criminating attribute, whereas for RAT, the critical factor is the dif- alarms to conjunction faces that were constructed from same-color par- ference in summed evidence across the two alternatives. In a training ents, regardless of whether those parents had been close or far on the environment in which ideal-observer TTB and RAT strategies yield study list. However, when compound words served as stimuli, no color- equivalent decisions, the response time results suggest that the vast based context effect was found. This indicates that context effects may majority of participants adopted the generalized TTB strategy. The re- be stimulus specific. The results disconfirm simple familiarity-based sponse time approach is also validated in an experimental condition explanations for some memory conjunction errors, and show that con- in which use of a RAT strategy was essentially forced upon subjects. text may serve as a retrieval cue for stimulus parts. 9:20–9:35 (9) 8:20–8:35 (12) Why You Think Milan Is Larger Than Modena: Neural Correlates Feature and Conjunction Errors for Unfamiliar Faces in a Con- of the Recognition Heuristic. KIRSTEN G. VOLZ, Max Planck Insti- tinuous Recognition Memory Task. TODD C. JONES, Victoria Uni- tute for Human Cognitive and Brain Sciences, LAEL J. SCHOOLER, versity, Wellington, & JAMES C. BARTLETT, University of Texas, Max Planck Institute for Human Development, RICARDA I. Dallas—In two recognition memory experiments, subjects viewed SCHUBOTZ, Max Planck Institute for Human Cognitive and Brain grayscale images of unfamiliar faces in a single list and judged the sta- Sciences, MARKUS RAAB, Flensburg Universität, GERD GIGEREN- tus of each face using a 6-point confidence rating scale (e.g., with 1 ϭ ZER, Max Planck Institute for Human Development, & D. YVES VON very sure new and 6 ϭ very sure old).