The experiments reported here illuminate current understanding of the factors influencing false recall. These experiments employ a procedure reported by Roediger and McDermott (1995), who demonstrated that reliable, predictable intrusions can be elicited in single trial free recall by presenting subjects with short lists of words (e.g., bed, rest, awake, tired, dream, etc.), all of which are associates of a critical nonpresented word (e.g., sleep); the critical nonpresented word is recalled and recognized as having occurred in the list. In Experiment 1, high levels of false recall were obtained, replicating Roediger and McDermott's (1995) results; in addition, although introduction of a short (30 s) filled delay resulted in forgetting of studied items, false recall was maintained over the delay. On a final free recall test occurring 2 days later, the probability of false recall was increased when an initial test had been taken on the first day, relative to a condition in which no initial test had occurred. In addition, the probability of false recall exceeded the probability of veridical recall on the final test. In Experiment 2, when multiple successive tests were taken, the probability that subjects would later claim to remember the critical nonpresented item (i.e., to recollect something specific about its presentation) was enhanced relative to a single test condition. Experiment 3 examined the persistence of this memory illusion by giving subjects multiple opportunities to hear the list (in a multitrial free recall procedure). Although the probability of free recall diminished across successive trials, robust levels of false recall were still obtained after 5 study-test trials. When subjects returned 1 day later, the probability of free recall was enhanced (relative to the last trial of Day 1), whereas the probability of veridical recall diminished across the delay. Overall, results demonstrate that illusory memories can be extremely robust, persisting across time and repeated retrieval attempts. In some cases, the probability of false recall exceeds the probability of veridical recall. Results converge upon the conclusion that the act of retrieval can play a critical role in determining false recall.

Discrete models of speaking maintain that semantic-syntactic and phonological representations are largely independent, whereas interactive accounts allow for mutual influence between them. The studies presented here investigated this issue by employing a task in which participants named pictures while instructed to ignore visually or auditorily presented distractor words. Previous results using this paradigm with auditory distractors have been used to support the discrete view (e.g., Schriefers, Meyer, & Levelt, 1991) whereas results with visual distractors have been used to argue for an interactive account (e.g., Starreveld & La Heilj, 1996a). The first two experiments served to clarify the discrepancy across distractor modalities. Experiment 1 demonstrated that with auditorily presented distractors semantic effects preceded phonological effects whereas with visual distractors phonological effects had an earlier onset than semantic effects. Experiment 2 provided a means for accounting for this discrepancy by demonstrating that the results for visual distractors followed the auditory pattern when presentation time was limited. The following two experiments addressed the issue of interactivity vs. modularity in speaking by employing auditory distractors and investigating the effects of complex types of distractors. Experiment 3 factorially crossed the factors of semantic and phonological relatedness by employing both semantically and phonologically related distractors (FLY-FLEA). An interaction between the two factors was obtained which was interpreted as supporting an interactive account of speaking. Experiment 4 investigated the effects of mediated distractors which are related to the picture name via an intervening word (TIDE-(TIGER)-LION). A potential effect of such distractors would require an interplay between semantic and phonological levels and thus further support an interactive view. The results showed no effects on naming latencies, a finding that probably is not diagnostic with regard to the question of interactive vs. modular accounts. The second part of this thesis introduced an interactive computational model closely related to Dell's (1986) model of speech production. This model yielded a sequence of semantic and phonological effects and showed an interaction between the two factors as well as the absence of mediated effects. In summary, the experiments and the model favor an interactive view of speaking in which semantic and phonological levels are closely interconnected.

Five experiments investigated whether phonological representations play a role in verbatim sentence repetition. Experiments 1 and 2 revealed that words related both semantically and phonologically to sentence words were more likely to be intruded into sentences than were words related only semantically. Experiment 4, the most stringently controlled, also yielded more intrusions for semantically and phonologically related lures, with a larger effect for abstract than concrete stimuli. Experiment 5 demonstrated that words related only phonologically to sentence words could be intruded into sentences as well. It is argued that veridical sentence recall taps phonological as well as semantic memory representations for accurate performance; a view inconsistent with the "Semantic Regeneration Hypothesis" offered by Potter and Lombardi (1990). Further, it is argued that the semantic and phonological representations act in an interactive fashion and support each other during sentence recall.

Brunswik's lens model (1955, 1956) was used in two experiments to provide feedback to subjects to use in improving baseball team win predictions. The lens model decomposes judgment into three component parts which were used as feedback; cue-criterion relationships (task information, TI), subject cue-utilization relationships (cognitive information, CI), and the relations between the criterion and subject predictions (functional validity information, FVI). In experiment 1, subjects given TI + CI or TI + CI + FVI improved their predictions to a greater extent than subjects given TI. It was also shown that subjects' prediction performance in the TI condition was moderated by domain knowledge. In experiment 2, subjects were unable to improve their performance in any of the lens model feedback conditions (TI, TI + CI, and modified CI). The type of domain knowledge termed "task knowledge" was found to be the most important predictor of performance.

The World Wide Web and search engines are widely used, and getting good results from searches is important. Research has shown that there are measurable, quantitative features of Web pages that relate to the quality of the Web page. There is little existing research that has examined a set of Web pages, rated by representative users, to determine what Web page features may predict users' ratings about the quality of the pages. This is the first study to apply policy capturing, a methodology to capture the cues people use in making judgments or ratings, to determine how a group of representative users made judgments about various Web pages. Search engines could utilize these features in examining Web pages to provide more useful results. The features of the Web pages examined were page length, links, images, keywords per page, and keywords per title. Users rated 40 Web pages on the relevancy, ease of understanding, and trustworthiness of the page in three separate topic areas. Analyses included the amount of variance accounted for, weighting of individual cues, complexity of the decision process, participants' insight into their own rating methodology, and the role of individual difference variables. Free response data were gathered about what other features influenced participants' ratings. Overall, average ratings on the three dependent variables ranged from 60--70 out of 100. The beta weights and R2 for the independent variables were low to moderate. Keywords were shown to be an important predictor in the page or title depending on the topic area. Links and images were more important for noninformation-centric topics. Most people used one of the five linear cues significantly, and more complex, nonlinear relationships were also found. For example, keywords increased the value of a page up to a point, but beyond that point the benefit decreased (a quadratic relationship). The amount of insight into how decisions were made about pages varied among participants from none to nearly perfect. Many of the variables and results varied significantly by topic area. Policy capturing is a good methodology for examining this issue, but additional Web page features should be tested.

In two experiments, subjects heard simple action statements (e.g., "break the toothpick") and, in some conditions, performed the action or imagined performing it. After 10 minutes, 24 hours, 1 week, or 2 weeks they imagined performing each action 0, 1, 3, or 5 times, where some actions had been previously heard and others had not. A later test showed that as the number of interpolated imaginings increased, overall recognition accuracy decreased in all conditions except when subjects imagined the events immediately before the test. Source monitoring was not systematically affected by the number of imaginings. However, imagination inflation was observed: As the number of imaginings increased, subjects increased their judgments that they had actually performed an event on the first occasion when they had not. Imagining events multiple times enhanced the hit rate for actions, and also increased subjects' claims that events had occurred when in fact they had not.

Gambling paradigms tapping both reward processing and decision-making tasks in control and patient populations have found differences in behavior based on individual differences in immediate reward representation. The current investigation examined decision-making in individuals who differed on self-reported measures of impulsivity and used event-related potentials (ERPs) to examine the network dynamics of reward and decision-making circuitry among low and high impulsive participants. An inferior frontal component, the anterior P2 (P2a), indexing orbitofrontal cortex (OFC) activity, and a medial frontal negativity (MFN), indexing anterior cingulate cortex (ACC) activity, were measured related to choices made from high-risk and low-risk decks of cards in two modified versions of the Iowa Gambling Task. Results indicated that the P2a indexed reward expectation in a single-presentation version of the Iowa Gambling Task and the MFN indexed evaluation of decisions in a dual-presentation version of the Iowa Gambling Task.

Two experiments were conducted to investigate product finding behavior on e-commerce Web sites, focusing on the use of product menus or site search functions to locate products. Experiment 1 sought to characterize typical product finding behavior on e-commerce sites, and to shed light on the user's decision to browse a product menu or use a site search function to locate products. The results yielded a wide distribution of searching and browsing behavior and found that use of the site search functions did not yield faster or more accurate performance in locating products. Questionnaire data suggested that perceived effectiveness of the menu structure, menu and search function prominence, and the user's disposition toward using search functions all influenced the decision of whether to browse or search a site for a product, with the principle known as "information scent" playing a particularly important role. Experiment 2 used experimentally controlled novel e-commerce sites to investigate the factors suggested to be important to product finding behavior by Experiment 1. The goal was to determine whether product finding behavior could be influenced via the manipulation of site design factors. The results provided evidence that such behavior could be manipulated through site design and supported the role of Experiment 1 factors in the decision process of the user. As a whole, the results suggest that product finding behavior is dependent upon both the characteristics of the site as well as the user. They also suggest that the efficiency of the menu structure of a site appears to be as critical to its usability as the fidelity of its search function. Further implications are also discussed.

Experiment 1 presents data from a patient with a semantic short-term memory deficit, patient ML, that indicate profound susceptibility to interference. For example, although this patient cannot reliably recall three items in a serial recall task, he paradoxically shows exaggerated effects of proactive interference in short-term memory. However, this patient's difficulty with interference appears to be limited to tasks involving verbal stimuli---other data show that patient ML performs normally on two nonverbal tasks that require resolution of interference. Experiment 2 attempted to identify ERP components related to proactive interference in one of the tasks administered in Experiment 1. This task, the recent negatives task, is a convenient measure of susceptibility to proactive interference. Moreover, Experiment 2 added an additional manipulation motivated by a unique effect discovered during the testing of patient ML (Hamilton, 2004), whereby the patient performed much better on the recent negatives task when repetition was minimized and the number of stimuli presented within the task were expanded. Two ERP components, a frontally distributed N400 effect and parietally distributed late positive component (LPC), were found to respond to the manipulation of recency and repetition. Experiment 3 and Experiment 4 examined a language comprehension paradigm known to differentiate between patients with semantic and phonological short-term memory deficits. This task requires detection of semantic anomalies in phrases in which multiple adjectives appear before or after a noun---multiple adjectives appearing before a noun are believed to place greater demands on semantic short-term memory relative to when adjectives appear after a noun. Thus, patients with semantic short-term memory deficits are especially poor at detecting anomalies in the before condition. Experiment 3 uses the parietally distributed N400 as an indirect measure of short-term memory demands to corroborate behavioral and patient data. Relative to the "after" condition, the "before" condition did elicit smaller N400s, consistent with the idea that integration of adjectives in the before condition differs from integration in the after condition. Experiment 4 uses a modification of the sentence anomaly task employed in Experiment 3 to identify brain areas engaged in short-term maintenance of semantic representations. It was hypothesized that the before condition would produce greater activation in the left inferior frontal gyrus, a region that has been related to short-term maintenance of semantic representations. Results from Experiment 4 are discussed in terms of the organization of maintenance and control processes important in semantic short-term memory. Finally, data from Experiments 1--4 are discussed in terms of their implications for theories of semantic short-term memory deficits and the associations of semantic short-term memory deficits with particular deficits of language comprehension and production.

Evidence from brain-damaged patients suggests that there are separate buffers for input and output phonological retention in verbal short-term memory (STM). This possible distinction was investigated with college students (Experiment 1 to 3) and deaf signers of American Sign Language (ASL) (Experiment 4) using different verbal materials in a serial probed recall paradigm. It is reasoned that natural linguistic input (speech for hearing people and ASL for deaf people) would be stored in an input phonological buffer whereas internally generated phonology derived from reading, naming pictured objects, or lip-reading would be stored in an output phonological buffer. In this study, participants were presented with memory lists in which presentation modality (spoken vs. lip-read word, written vs. lip-read word, etc.) was changed after every second item. A probe item from the list was repeated at the end of the list and participants were instructed to either recall the item in the list that has immediately followed the probe or recall the first item after the probe that is in the same modality. Some of these same-modality items were temporally distant, that is, having two intervening items of a different modality. It is predicted that the temporally distant probe in the same modality with the target results in higher memory performance than the temporally adjacent probe in a different modality only if the switch in modalities is between input and output phonological forms. The results from Experiment 1 demonstrated that spoken words and written words were stored in the input and output phonological buffers, respectively. The results from Experiment 2 and 3 further supported the hypothesis in showing that written words were retained in the same buffer with lip-read words and with nameable pictures, while spoken words were retained in a different buffer from these materials. The findings from lists consisting of words in ASL and nameable pictures in Experiment 4 were not conclusive. However, preliminary data suggested that there might also be a separation between signed words and nameable pictures. Overall, the findings from this study conformed to the predictions from the hypothesis of separate input and output phonological retention.