Abstract not available.

The relation between mental ability and speed of auditory discrimination was investigated during an auditory oddball task with backward masking. Behavioural and electrophysiological data were collected from 58 females. Across target discrimination conditions that varied in the interval between the target and the masking stimuli and in the tonal frequency of the target and masking stimuli, HA participants displayed faster RT and more accurate discriminations than LA participants. HA participants also had shorter P300 and MMN latency and larger P300 amplitude than LA participants. The effects suggest that the speed of accessing STM is faster for HA than LA participants. Moreover, the pattern of results obtained with these data eschews task difficulty effects that would endorse a sensory discrimination hypothesis.

Theories of contingency judgements generally agree that (1) memory is a structure that possesses a limited capacity and that (2) it plays an important role in the detection and assessment of covariations. Empirical evidence, although limited in the specific context of contingency judgements, seems to support these notions. While theorists agree that some information needs to be held in memory in order to reach a contingency judgement, they disagree, however, on the exact type of information. As a result, they offer different predictions as to what would increase memory load as involved in contingency judgements. Kareev (1995, 1997) implicitly assumes that people attempt to memorize the sequence of events leading to a contingency judgement and, therefore, the longer the series of events, the higher the memory load. On the other hand, Wagner (1976, 1981; Rescorla & Wagner, 1972) proposes that people base their judgement on the strength of a "mental bond" rather than the recall of the series of episodes. In this view, the manipulation that taxes memory capacity is not the length of a series of events but the presence of multiple simultaneous contingencies. The current thesis aimed at clarifying the role that memory plays in the assessment of covariations by contrasting these two opposing viewpoints. Five experiments examined the role of memory capacity in contingency judgements by means of: (1) increasing the length of the series of single events experienced; (2) increasing the number of contingencies presented simultaneously; (3) examining the effect of individual memory capacity. Results generally support Wagner's theory with additional findings falling outside of the theory's explanatory power.

In 4 experiments, participants judged the contingency between compound predictors and an outcome, as well as the contingency between the compounds' constituent elements and the outcome, in different contingency and similarity conditions. The Rescorla-Wagner (1972) and Pearce (1987) models of associative learning describe different processes through which a compound predictor becomes associated with an outcome, and how responses to a compound are mediated by the association between its constituent elements and the outcome. According to the Rescorla-Wagner model, when a compound is paired with an outcome an association will develop between each element of the compound and the outcome, and responding to the compound will reflect the associative strength accrued to each element in an additive fashion. According to the Pearce model, a compound is associated with the outcome in its entirety (i.e., as a configural cue), and responding to the compound is related to both the associative strength of the configural cue and the associative strength generalised to the compound from other predictors as a function of similarity. Across experiments and conditions, compound predictors were assessed independently of the normative relation between their constituent elements and an outcome. Manipulations of the similarity among predictors, measured as the proportion of elements they share, did not impact judgements of compound predictors. Findings are consistent with the notion that compound predictors are functionally independent of their constituent elements, and possible modifications of the Rescorla-Wagner and Pearce models to account for these findings are discussed. A configural associative model that assumes no generalisation of associative strength among predictors appears to provide the best fit to the empirical findings.

The cognitive approach to the study of emotion is grounded in the belief that individuals are active agents and are able to exercise control over thought processes, motivation and behavior. Cognitions, causal thoughts in particular, play a central role in behavior and affect generation. Emotion depends upon how the individual cognitively appraises an event, not the event per se. Vallerand (1987) has proposed an intuitive-reflective appraisal model for self-related affects to examine the roles of intuitive (e.g. subjective performance assessment) and reflective (e.g. causal attributions) appraisals in the generation of self-related and general-type affects. The purpose of the present study was to test this model in a competitive fencing tournament and ascertain the relationships between three cognitive antecedents (intuitive appraisal, causal attributions, and intellectualization) and self-related and general-type affects. Results showed support for the main postulates of the model. Intuitive appraisal was found to have important and necessary effects on self-and general-type affects. This intuitive appraisal was shown to have more of an effect on affects than objective outcome (win/loss). Reflective appraisal processes, in the forms of causal attributions and task importance, were not necessary for affect generation, though task importance played a significant minimizing role in the experience of positive self- and general-type affects, particularly in the perceived failure condition. The results of the present study promote the use of sport specific models in the understanding of the relationship between athletes' cognitive antecedents and affect generation.

Lexical decision and pronunciation tasks were used to investigate semantic priming, the finding that a word is quicker to recognize when it is preceded by a related than an unrelated stimulus. The first experiment involved a lexical decision task (LDT) in which skilled and less-skilled readers made decisions about letter strings that were preceded by conceptually-related or unrelated stimuli. The effects of time to process the prime and type of prime (word or picture) were examined. Word and picture priming effects were observed at short and long time intervals with skilled and less-skilled groups. Finding word priming was not surprising; there are prior studies that have documented priming by words. However, prior experiments on picture priming have methodological flaws such as multiple presentations of stimuli that make it unclear whether pictures can prime word targets through semantic and nonstrategic routes. The facilitation of word targets following word primes may be due to semantic relations as well as associative relations. While picture priming provides evidence of semantic priming, picture priming cannot be associative at a lexical level because no orthographic features are displayed. During the processing of a picture, information about the picture and its related concepts are activated which facilitates processing of a subsequently presented word. The effect of priming was greater with picture primes than word primes, perhaps because the associations were stronger between the picture-word pairs than the word-word pairs. Because priming in a LDT may be attributed to postlexical checking, priming was further investigated in a pronunciation task in which the strategy is not helpful. In Experiment 2, in which subjects pronounced words that were preceded by related or unrelated pictures, the priming effect was significant. Finding a picture priming effect is important; it supports the interactive view that pictures provide a context that affects the processes that occur before word recognition. Priming must be due to semantic associations between the picture prime and the words corresponding to related concepts. Word and picture priming can be explained by current models of lexical access.

This study investigates a phenomenon that I have termed occlusion inhibition. This research and a small number of earlier studies suggest that, in some experimental conditions, when an attended (target) object is partially occluded by a distractor object, there is less attention allocated to the occluded region of the target object than to the visible parts of that object. In the literature, there are mixed results concerning this attentional effect. Some studies find it and others do not. This study investigates the differences between those conflicting studies with the goal of identifying the factor or factors that govern when occlusion inhibition occurs. Evidence is presented to rule out a number of potentially relevant factors such as depth perception, figural complexity, set size, the use of real world vs. abstract geometric objects, the position of occlusion, the number of overlaps in the display, and the adoption of the attend-object paradigm over the spatial cueing paradigm. After all these factors are ruled out, Experiments 3 and 4 provide evidence for a factor that does determine whether occlusion inhibition occurs or not. These two experiments differ only in the fact that participants are required to report the border color of the target object in Experiment 3 and not in Experiment 4. This task was designed to ensure that participants fully attend to the target object. Occlusion inhibition occurs when the target color is reported, but not when no target color report is required. Removing the target reporting task was found to be an effective means of turning occlusion inhibition on and off. The results of these experiments suggest that, if occlusion inhibition is to take place, attentional selection of overlapping figures requires the target object to be fully processed. This conclusion in turn suggests that attention does not automatically exclude the irrelevant portions of occluded objects, but that attention selects the entire location of the object and then, through reiterative feedback mechanisms, fine tunes the information to inhibit areas that do not belong to the object.

Is letter information processed in parallel or serially when readers encounter words? Kwantes and Mewhort (1999a; 1999b) addressed this question by introducing the concept of a word's orthographic uniqueness point (OUP). A word's OUP refers to the letter position in a word where that letter pattern uniquely identifies the word. They found faster naming times for words with an early OUP versus words with a late OUP. They argued that the faster naming times for the early OUP words is inconsistent with parallel implementations of models of letter processing. However, the overall letter overlap between the target word and other words that a reader knows was not controlled (Lamberts, 2003) and might account for the differences in naming times rather than for differences in the position of a word's OUP. Readers might be sensitive to any difference in the degree of letter overlap with other words and parallel letter processing accounts could be sensitive to this difference. Two experiments were conducted to extend OUP findings to a more natural reading context by monitoring eye movements while readers read sentences. The first experiment used stimuli with a relatively large difference in the position of a word's OUP. The second experiment matched the early and late OUP words in terms of the overall letter overlap with other words and unconfounded the initial trigram frequency. In both experiments, readers did not benefit from a word possessing an early OUP. The results were discussed in terms of potential serial and parallel letter processing accounts.

The purpose of the dissertation was to investigate how we store and access affixed English words by monitoring participants' eye movements during sentence reading. Previous research (Niswander, Pollatsek & Rayner, 2000) indicated that both the frequency of the root morpheme and the frequency of the whole word affect the fixation time on a suffixed word while reading. The current research findings confirm the effects of both root and whole-word frequencies for prefixed words in English. In addition, the current research indicates that word length plays an important mediating role: for prefixed words, there is clear evidence that for long words the effect of root frequency is dominant, whereas for short words, the effect of word frequency is dominant. This effect mirrors a finding in Finnish (Bertram & Hyönä, 2003) for compound words In contrast, the word length by frequency results were less clear for the suffixed word sets. Finally, the effect of affixal homonymy was investigated. Reading times on words containing homonymic affixes were compared to words whose affixes were not homonymic, and there was some evidence that a homonymic affix interferes in the processing of the word. Models of complex word processing were discussed in the context of the research findings.

Visual search consists of locating a known target amongst a field of distractors. Often times, observers must search for more than one object at once. Eye movements were monitored in a series of visual search experiments examining search efficiency and how color is represented in order to guide search for multiple targets. The results demonstrated that observers were very color selective when searching for a single color. However, when searching for two colors at once, the degree of similarity between the two target colors had varying effects on fixation patterns. Search for two very similar colors was almost as efficient as search for a single color. As this similarity between the targets deceased, search efficiency suffered, resulting in more fixations on objects dissimilar to both targets. In terms of representation, the results suggest that the guiding template or templates prevailed throughout search, and were relatively unaffected by the objects encountered. Fixation patterns revealed that two similarly colored objects may be represented as a single, unitary range containing the target colors as well as the colors in between in color space. As the degree of similarity between the targets decreased, the two targets were more likely to be represented as discrete separate templates.