Acting according to one's own internal goals is crucial to flexible behaviour. Clinical and lesion studies of patients with frontal lobe damage have demonstrated syndromes potentially resulting from deficits in the cognitive control system for internally-guided behaviours. Some patients can perform well on tasks that are well-constrained by the environment, including standard measures of IQ, yet show impairments in everyday life and laboratory equivalent ill-structured tasks that make planning and self-cueing demands. This thesis is concerned with the executive control of such tasks in the healthy population. Eight experimental studies are reported which consider the role of endogenous and exogenous cueing in prospective memory (PM) and multitasking. Experiments 1-4 integrated the two standard laboratory-based paradigms of task-switching and PM to assess the independence of processes involved in externally-cued task-switching and self-initiated (i.e. internally-generated) PM task switches. These experiments suggested that these two types of task switches are enabled by independent processes. Focusing only on PM, Experiments 5-6 manipulated the degree of internal cueing required by the PM task and analysed the effects on performance of the ongoing task. Participants exhibited poorer ongoing task performance in a time-based PM task without the presence of a clock (internally-cued) compared with PM tasks with stronger external cues (with a clock and event-based). The results support the view that the executive processes recruited for PM tasks reflect the demands made on internal control. In Experiments 7-8, individual * differences in internally-guided control processes were explored after development of an advanced multitasking test (AMT) for the healthy population. AMT performance correlated with some real-life outcome measures. The evidence in this thesis supports the suggestion that different executive processes are employed depending on the demand for internally- generated behaviour. Individual variation in the cognitive control system for internally- guided behaviour may relate to everyday functioning.

No description available.

How do we represent our world and how do we use these representations to reason about it? The three studies reported in this thesis explored different aspects of the answer to this question. Even though these investigations offered diverse angles, they all originated from the same psychological theory of representation and reasoning. This is the idea that people represent the world and reason about it by constructing dynamic qualitative causal networks. The first study investigated how mock jurors represent criminal evidence and reason with such representations. The second study examined how people represent the causes of a complex environmental problem and how their individual representations are directly linked to how they reason about the issue. The third and final study inspected how people represent causal loops and reason in accordance with these cyclical representations. These studies suggest that people do represent the world by arranging evidence, causes, or pieces of information into a causal network. In addition, the studies support the idea that these networks are of a qualitative nature. All three studies also indicated that people update their representations in accordance to a dynamic world. The studies specifically explored how reasoning, and therefore judgment is linked to these representations. The thesis discusses the theoretical implications of these and other findings for the causal model framework as well as for cognitive science more generally. Related practical implications include the importance of understanding naïve causal models for applied fields such as legal decision-making and environmental psychology.

The recent trend within cognitive science toward the adoption of a de re account of semantic content has been accepted by some theorists without due consideration, an oversight that has generated a number of logical inconsistencies in the literature. This thesis endeavours to shed light on this omission, and the effects thereof, by scrutinising the account of one such theorist: The psychologist Philip Johnson-Laird. Herein, Johnson-Laird’s introduction of a Peircean iconism is subjected to a thorough analysis and, in this way, an important example of the problematic tensions created by a failure to address the substantial nature of a theory, before its wholesale integration into a philosophy, is revealed. The philosophic method of analysis and synthesis is applied to the arguments and hypotheses of Philip Johnson-Laird, thereby demonstrating the base assumptions that constitute his account of cognition and the relationships between these assumptions. Investigation of the results of this method identifies logical inconsistency present in the hypothesis and, following Popper and Quine, logical inconsistency is taken as a disproof of a theory. The study concludes that, although Johnson-Laird’s updated philosophy was able to dissolve historical arguments against his theory, his use of Peircean iconism is inconsistent in its own right and produces serious tensions with aspects of his extant philosophy. As such, it is concluded that Johnson-Laird’s externalist semantics fails and it is recommended that his account of semantics be reconsidered.

No description available.

No description available.

The work presented in this thesis aimed to advance an understanding of belief bias in human reasoning. What are the factors that cause people, when engaged in reasoning, to be influenced by their prior beliefs and knowledge rather than apply logic to an argument to arrive at a normatively correct outcome? To address these issues three experiments were undertaken that focused on cognitive ability measures (specifically working memory capacity) as well as evidence from response time data and individual differences in measures of thinking dispositions. Experiment 1 examined the role of working memory in the belief bias paradigm, considering both endorsement rates and processing times. Experiment 2 built on the findings from Experiment I and investigated more thoroughly the role of working memory capacity and the evidence from the chronometric data. This was achieved by using materials with stricter controls and through increased test power via a larger participant sample. Experiment 3 aimed to ascertain exactly what it is that drives successful reasoning with belief-oriented problems. To address this issue the experiment involved a measure of abstract reasoning as well as measures of thinking dispositions, one assessing people's proclivity to engage in open-minded thinking and the other assessing their tendency towards rational and experiential thinking styles. Overall the findings suggest that reasoners who apply open-minded thinking have a greater ability to override the effects of belief bias and perform in a more normative manner when logic and belief are in conflict. A higher working memory capacity is only associated with reasoners demonstrating longer processing times with conflict problems, but plays no role in successful reasoning in the belief bias paradigm. With respect to processing times across problem types, the data revealed some support and some challenges to previous models of belief bias.

In a groundbreaking empirical study Onishi and Baillargeon (2005) developed a nonverbal version of Wimmer and Perner's (1983) unexpected transfer false belief task based on Woodward's (1998) violation-of-expectation methodology in order to assess 15-month-olds preferential looking times. The findings from this study demonstrate that infants looked less when an agent reached for a target object where the agent last . . . ..- saw it when it had been moved to a different location in her absence. Onishi & Baillargeon and others have interpreted these data as providing the first empirical demonstration of an implicit and innate meta-representational understanding of beliefstates within infancy (the rich accounts). Alternatively, these data have been interpreted at a mentalistic level as demonstrating no more than infants' understanding of the agent's ignorance (Southgate et aI, 2007; Wellman, in press) or at a basic non-mentalistic level involving an understanding of actions based on associationist strategies and behavioural rules (the lean accounts of Perner & Ruffman, 2005). I suggest that the rich and lean accounts are problematic in that both provide a non-developmental and excessively adult-like conception of task performance on this measure. This masks the emergence of, as yet unconsidered, lower level and developmentally sensitive precursor competences. I argue that infant false belief data can be better understood if situated within such a lower level developmental framework which emphasises (a) infants' understanding of goal-directed actions and (b) key differences between the levels of perceptual awareness demonstrated by infants and older children in the verbal and non-verbal versions of the unexpected transfer task. At a wider level I argue that the types of competence demonstrated on both verbal and non-verbal false belief tasks can be situated within a second person I .~------------ -- r relational framework (e.g., Gallagher, 2005; Reddy, 2003 , 2008). With specific reference to Onishi and Baillargeon's (2005) non-verbal 'false belief' task I argue that infant social competence is framed in terms of what 'we' saw (a first-person plural explanation) rather than what 'I' saw 'you' see (a third-person explanation). In order to test these assertions five cross-sectional empirical studies were conducted on infants between the ages of 10- and 22-months. The findings from at least four of these studies (Studies 1-, 2-, 3-, and 5) were inconsistent with both the rich and lean accounts. In contrast, the findings from at least three of these studies (Studies 1-, 3-, and 5) were consistent with my own lower level action-perception account. These findings suggest a reconceptualisation both in terms of how we assess and how we interpret infant false belief data.

Both the production of knowledge and the product, knowledge itself, are social phenomena. This generally accepted fact is generally thought to require relativism, scepticism, and Kuhnian incommensurability, as well as casting serious doubt on the potential of our cognitive traditions to provide us with objective knowledge about an objective world. This thesis exposes and critiques the presuppositions about the nature of reasoning and objectivity which underlie these fears. Combining a Nietzschean, perspectivist account of objectivity with a conception of reasoning drawn from Lockean epistemology and pedagogy, I build a new account of cognitive optimality, dubbed 'Linkmaking'. The phrase deliberately encompasses several meanings. We 'make links' by noticing connections between objects in the world, by linking ideas together to form a theory or a curriculum; by forming social connections, and by developing interdisciplinary practices. I defend the view that we cannot fully address any of these kinds of Link without reference to all of the others. I further show that out best means to critically assess our cognitive groups is to evaluate the extent to which those groups encourage Linkmaking practices. The major potential challenge to Linkmaking is Kuhnian incommensurability. Having demonstrated the flaws inherent in Kuhn's account, this thesis defends the weaker, Doppeltian form of incommensurability, which grants us insight into the genuine problems which can occur in interdisciplinary research. We then see that the Strong Programme in the sociology of knowledge, inspired by the strong, relativistic version of the Kuhnian incommensurability thesis, has held sway among sociologists because they do not generally study interdisciplinary practices, which highlight scientists' (perspectivist) objectivity. Furthermore, social scientists who accept Kuhnian constructivism doubt their own potential for objectivity, presuming the presence of strong incommensurability where there is none. Undertaking Linkmaking practices both cures this illusion, and improves the cognitive optimality of the group.

The aim of the research presented in this thesis was to investigate the effects of trainin- on reasoning and decision making performance. In Experiment Ia study is reported which examined the relationships between performance on a variety of reasoning tasks and measures of individual differences. Tasks employed were documented in the literature for their differential responding according to heuristic and analytic processes. The reasoning tasks to be utilised in the training studies were also validated. In Chapter 4, two statistical training studies are reported which demonstrate that analytic responding on everyday reasoning problems can be increased after instruction on the Law Of Large Numbers. Bias was eliminated, but only on written justifications of their responses. Belief-based responding was still utilised when participants were asked for a quick indication of argument strength on a rating scale. This demonstrates a dissociation between analytic and belief-based responding. A second series of experiments explored the effects of both abstract and schema-based training on selection task responding. All the training procedures resulted in positive transfer apart from training on the logic of the material conditional which facilitated perforinance on arbitrary tasks only. Relationships between perforinance on the tasks post-training and cognitive ability indicated that training was more effective for higher ability participants. The differential training effects were discussed in terrns of complexity of training procedures. The findings overall have implications for dual process theories of reasoning. The findings suggest that the interaction between training and System I and System 2 tasks/responses is a great deal more complicated than the simple analysis that is afforded by dual process accounts.