Keeping up with current trends in cognition

Mather, George (2005) Keeping up with current trends in cognition. Trends in Cognitive Sciences, 9 (9). pp. 414-415. ISSN 1364-6613

Full content URL:

Full text not available from this repository.

Item Type:Review
Item Status:Live Archive


The modern Oracle, otherwise known as Google, provides a number of definitions of the term ‘cognition’. Most definitions refer to the mental functions and/or neural processes that underlie knowledge-based thought – including attention, language, memory, perception, and reasoning. The cover of this edited book declares that its aim is to offer a guided tour of up-to-date research in all these areas. It is intended for advanced students and researchers in related areas who have a general grounding in the subject but need an update in less familiar areas of cognition.

By and large, the book makes a very good stab at achieving its aim. The chapters divide into three groups. First, there are fourteen chapters devoted to specific areas of cognition, sub-divided into perception/attention/action, learning/memory, language, and reasoning/decision making. Then there are three chapters on cognitive neuropsychology, covering object recognition/action, learning/memory, and language. Finally there are two chapters on modelling. The authors are well-established, authoritative figures in their respective fields, giving the reader some confidence regarding their chapter contents.

The edited form of the book means that chapters are relatively independent, so it is possible to read the coverage of a specific issue without the assumption that previous chapters need to have been read. Inevitably, the authors present a personal view of the field, emphasizing developments that they perceive as particularly important or interesting. Take, for instance, the very first chapter on Visual Perception by Wagemans, Wichmann and Op de Beeck. It offers an excellent perspective on current trends in the area, assessing the utility of the traditional ‘measurement approach’ to psychophysics (which assumes that early processes take measurements of the image). It also includes extended discussions of Gestalt and Gibsonian approaches. The latter might not figure highly on everyone's list of ‘hot topics’, but are welcome nevertheless. Humphreys and Riddock's chapter in the cognitive neuropsychology section draws heavily on their own work, but I guess there is a relatively limited population of relevant neuropsychological patients, and they have probably worked with most of them. It is reasonable nevertheless to expect the reader, who is likely to be a grown-up researcher, to take the personal nature of each chapter in their stride.

The two chapters on modelling are particularly welcome. Formal models have become an integral and indispensable part of research in all areas of cognition, from early sensory processing to semantic processing. Indeed, examples can be found throughout the book. Yet it could be argued that psychologists have been relatively slow to pick up the tools developed in the physical sciences for creating and evaluating models rigorously. Koen Lamberts's chapter provides a basic introduction to mathematical models, covering model types and parameters, and goodness-of-fit estimation. Myung, Pitt and Kim's chapter moves up a level to consider model specification and evaluation in more detail. Model evaluation is a recurring issue in research papers. If several models are available to explain a particular research finding, which is to be preferred? Myung et al. briefly describe some criteria. They consider the most important to be ‘generalizability’; a model's ability to fit not just the data already observed, but future data samples taken from the same process. A model's ability to fit a specific set of data is determined jointly by its ability to capture the underlying process and its ability to accommodate the noise that is inevitably present in specific real-world behavioural measures. These two factors need to be disentangled in order to judge the adequacy of the model. Complex models with many degrees of freedom absorb noise easily, improving fit regardless of their ability to capture the process, so goodness of fit alone is not an adequate measure of a model's success. Myung et al. discuss mathematical methods to assess generalizability, particularly Minimum Description Length. This technique, borrowed from computer science (other material is available via the Oracle, such as:∼pdg/ftp/mdlintro.pdf), is a formalization of Occam's Razor, according to which the best hypothesis for a given set of data is the one that leads to the largest compression of the data, or makes the minimum number of assumptions.

Overall, The Handbook of Cognition has sufficient depth to be one of the first ports of call for researchers who want to be brought up to speed on recent issues in cognitive psychological research.

Subjects:C Biological Sciences > C850 Cognitive Psychology
Divisions:College of Social Science > School of Psychology
ID Code:17020
Deposited On:08 Apr 2015 10:49

Repository Staff Only: item control page