E-viri
Odprti dostop
-
Milovanović Goran
09/2013Dissertation
The subject of the present meta-theoretical and methodological analysis is the scientific status of the concept of rationality of cognition in Computational Cognitive Psychology (CCP). On the first level of analysis rationality is constrained as a subject matter of cognitive psychology, and only on this level of analysis it is possible to reach a conclusion on whether a cognitive function is rational or not. This empirical question is characteristic of the contemporary rationality debate. On the second level of analysis, rationality stands as a theoretical and methodological concept which is built a priori in the foundations of the contemporary CCP. The central goal in this thesis is to clarify the relationship between these two concepts of rationality . In Part I we introduce the distinction between (i ) rationality constrained as a subject matter of CCP and (ii ) rationality constrained as a framework for theoretical and methodological decisions a priori in CCP. Through conceptual and diachronic analysis of the development of decision theories in social sciences we demonstrate the development of the contemporary opposition between normative and descriptive theories in cognitive science. We introduce Anderson's rational analysis as a central methodological assumption of rational theories and provide a short overview of the discussions that follow. In Part II we provide a complete critical analysis of the theoretical structure of CCP as a natural science of cognitive functions. We first discuss the foundations of the behavioural approach to the measurement of internal constructs, which presents the central theoretical problem of scientific psychology in general, and demonstrate the way that this problem is related to the problem of the axiomatization of decision theories. Then we provide a detailed discussion of CCP in three actual theoretical paradigms: symbolicistic, connectionist-emergentistic and constructivist-enactivistic. We discuss the typology of scientific explanation involved in these theoretical paradigms. We demonstrate that some parts of the program of the standard symbolicistic CCP cannot be falsified by means of standard behavioural methodology. Finally, we define a meta-theoretical framework for the analysis of formal cognitive theories: a set of theoretical concepts upon which the positions of v rational theories and theories of bounded rationality are developed in the rationality debate. In Part III of the discussion we present a meta-theoretical analysis of five groups of formal theories in contemporary CCP: theories of choice under risk and uncertainty, theories of causal learning, theories of memory, theories of judgement and reasoning, and theories of concepts and categorization. The results of our discussion uncover new, previously unstudied problems in the rational analysis of cognitive functions. Our analyses suggest that it is di cult, if not impossible, to assert that cognitive systems formulate unique goals of cognitive computations in front of uniquely defined problems of adaptation, which is a basic (implicit) assumption of rational analysis. This question naturally connects to the questions of (i ) the existence of the representative subject of any formal cognitive model and of (ii ) the possibility of parallel explanation of same behavioural data by multiple theoretical models. The analysis of the relationship between the concepts of rationality, adaptation and optimization, shows that it is di cult or impossible to differentiate between rationality and optimality in contemporary rational analyses, which complicates additionally the question of the adequacy of normative frameworks. We also discuss the status of intuitive justification of the axiomatic framework of contemporary descriptive theory of choice. In Part IV of the thesis we present a short overview of the history of ideas that led to the development of the contemporary CCP. We recognize three historical lines that converge in the development of CCP in the second half of the XXth century. The first relates to the tendency to provide a mechanicistic account of human mental processes and dates back to the philosophy of Descartes. This line reaches its peak in the discussion of the philosophy of mathematics between the formalists and intuitionists - the discussion that brings the key theoretical concepts of CCP as its consequences, such as the concept of computation, or the concept of a formal system. The second historical line can be recognized in the continuity of the development of probability theory with the study of the problems of rational choice and judgement under risk and uncertainty in contemporary CCP. The beginning of the third historical line that converges in the contemporary CCP is in the discovery of thermodynamics and statistical physics in the second half of the XIXth century, and it can be recognized in contemporary CCP as an impulse to constrain cognitive problems as optimization problems. In Part V of the thesis we develop a formal, mathematical model of one vi rational, Bayesian decision theory. It is our goal to demonstrate the possibility of a decision theory in which the cognitive processes responsible for deviations from normative decisions stand in isolation from the cognitive processes involved in pure choice. As a decision model we use Viscusi's Prospective Reference Theory (Viscusi, 1989), in which we then incorporate an alternative model of belief formation. The resulting Confidence Theory is able to explain all standard, robust deviations from normative decisions. We present experimental tests of the basic assumptions of the Confidence Theory (experiments 1a and 1b), as well as two attempts at model selection, through experimental measurement of certainty equivalents (experiments 2a and 2b) and choice experiments (experiments 3a and 3b). Confidence Theory explains empirical choices slightly better than the Tversky and Kahneman's Cumulative Prospect Theory (Tversky & Kahneman, 1992) while encompassing some empirical effects that Cumulative Prospect Theory cannot incorporate even in principle. The analysis of the homogeneity of preferences shows that this important condition is not satisfied in empirical choices, which implies that the canonical parametric form of the Cumulative Prospect Theory is not correct. We compare the theoretical, explanatory structures of these two decision models (which we use as representatives of rational and boundedly rational theories in general) and formulate the theoretically important distinction between dispositional and representational cognitive theories. This distinction is then used in Part VI to formulate one of our main arguments in this thesis. Part VI presents the key arguments that we use to support the following general conclusion: the rationality of cognition is a concept that should be excluded from the scientific discussion of cognitive functions. The first argument is based on the analysis of the possibility that there is always a representational cognitive theory that matches the behavioural predictions of the respective dispositional cognitive theory. If the former statement is true, and we argue that it is very probably true, than it does not make any sense to characterize any observable behaviour as rational or boundedly rational. The decision of whether cognition is rational or not reflects merely a choice between two equivalent languages of scientific description. Our second argument is based on the problem of choice of uniquely defined goals of cognitive computations that was recognized in Part III. If a cognitive system does not represent the problem of adaptation as a unique computational problem, rather interpreting the source of environmental information as demanding different, parallel cognitive computations, it is optimal for such a system to solve vii the adaptation problem by adopting a mixed-strategy as defined in game theory. Accordingly, every cognitive act is always an act of strategic interaction with the environment. To be able to discuss this situation in exact terms, we develop a strategic view of the cognitive system, in which cognitive acts take the form of mixed strategies: probability distributions defined on the space of formal cognitive models. We show then that a cognitive system which approaches the problem of adaptation strategically leaves a behavioural trace upon which it is impossible to decide whether it was produced by a rational or a boundedly rational system. We discuss several examples to support this observation. In the analysis of strategic interactions between different cognitive systems we also show that it is impossible to separate the question of the rationality of cognition from the question of the selection of goals that the respective cognitive functions are expected to satisfy. Finally, our third argument continues the discussion of creative conceptual functions that began in Part III. This argument shows that from the strategic interpretation of cognitive functions it follows that is not possible to formulate normative standards for the processes of symbolic interpretation. It is shown that just the opposite of stable cognitive strategies is what is necessary to support the conventional, arbitrary nature of symbols - that in turn enables for a potentially infinite number of symbolic interpretations. If cognitive strategies in symbolic interpretations remain fixed, the fundamental conditions of arbitrariness and conventionality are not met. In the domain of symbolic cognitive functions, thus, rational analysis is not possible. In the last part of the thesis we provide a more concise overview of our results and reiterate the general conclusion reached in Part VI: rationality of cognition is not a concept of a scientific theory, and does not deserve a scientific treatment in the scope of the discourse of computational cognitive psychology. Predmet metateorijske i metodološke analize u ovoj tezi je
Avtor
![loading ... loading ...](themes/default/img/ajax-loading.gif)
Vnos na polico
Trajna povezava
- URL:
Faktor vpliva
Dostop do baze podatkov JCR je dovoljen samo uporabnikom iz Slovenije. Vaš trenutni IP-naslov ni na seznamu dovoljenih za dostop, zato je potrebna avtentikacija z ustreznim računom AAI.
Leto | Faktor vpliva | Izdaja | Kategorija | Razvrstitev | ||||
---|---|---|---|---|---|---|---|---|
JCR | SNIP | JCR | SNIP | JCR | SNIP | JCR | SNIP |
Baze podatkov, v katerih je revija indeksirana
Ime baze podatkov | Področje | Leto |
---|
Povezave do osebnih bibliografij avtorjev | Povezave do podatkov o raziskovalcih v sistemu SICRIS |
---|
Vir: Osebne bibliografije
in: SICRIS
To gradivo vam je dostopno v celotnem besedilu. Če kljub temu želite naročiti gradivo, kliknite gumb Nadaljuj.