Theoretical Approaches To Reasoning

Deductive reasoning research covers a wide variety of tasks from syllogistic reasoning, to reasoning with spatial connectives, to reasoning with propositional connectives (e.g., if, or, and not). Any adequate theory of deduction should be able to explain the phenomena arising from this research. At this point there are probably only two real candidate theories to meet this challenge (the abstract rule and mental models theories), although there are other accounts that cover smaller sets of phenomena. Each of these are briefly outlined next and then some of them are expanded on in subsequent sections.

Abstract rule theories of deduction

The abstract rule theory generally takes logical notions of validity as its normative model of reasoning (see previous section). It assumes that people reason validly by applying abstract, content-free rules of inference, in a manner that is similar to the derivation of proofs in logic. In short, people employ a form of mental logic to derive conclusions from premises. However, people can make mistakes because some derivations are more complex than others (and exceed working memory) or because they misunderstand the premises of a given deductive problem. The main proponents of this view are Braine and O'Brien (1991; Braine, 1990; O'Brien, 1993, 1995; O'Brien, Braine, & Yang, 1994) and Rips (1994).

Mental models theory

The mental models theory also, in essence, assumes logical notions of validity as its normative model (Johnson-Laird, 1999; Johnson-Laird & Byrne, 1991). It assumes that people reason by manipulating mental models of a set of premises, in a manner akin to semantic methods of proof in logic. In short, people construct mental models that represent possible states-of-affairs in the world, and then they describe and verify these models to reach valid conclusions. A conclusion is valid if there are no counterexamples to it; that is, if there is no state of affairs in which the premises are true but the conclusion is false. Again, however, people may make mistakes if they have to represent a large number of models that exceed their working memory. The main exponents of this view are Johnson-Laird and Byrne (1991, 1996; Johnson-Laird, 1983, 1995a, b, 1999).

Domain-specific rule theories

Most domain-specific rule theories are essentially dual-process theories. They assume that some basic logical competence is handled by some core mechanism (be it an abstract rule or mental models one), but that there is a second mechanism using domain-specific rules that handles certain effects. Thus, reasoning is, in part, based on rules that are sensitive to the content of different situations; rules that are encoded in domain-specific schemata (see Chapter 9). There is a wide variety of such theories that propose different flavours of rules from pragmatic reasoning schemata (Cheng & Holyoak, 1985; Cheng, Holyoak, Nisbett, &

Oliver, 1986; Politzer & Nguyen-Xuan, 1992) to social exchange schemata (Cosmides, 1989; Cosmides & Tooby, 1992).

Heuristics and bias accounts

Most heuristic/bias accounts are also essentially dual-process theories. They assume that people have a basic logical competence, which is sometimes over-ridden by various heuristics or biases. Reasoning is seen as being, in part, due to non-logical tendencies based on a response to superficial aspects of a task situation (e.g., the presence of matching negatives, the position of an item on a test screen). We have called these accounts rather than theories, as they are a loose collection of ideas rather than a coherent theory. They also carry a theoretical health warning in that they can involve reified phenomena as cognitive processes; that is, the tendency to turn the description of a phenomenon into a theory. For example, one might find in a particular reasoning problem that people always choose the conclusion on the top left of the screen and then "explain" this by saying that people have a left-bias heuristic, when we really should be seeking a deeper explanation of why such a bias might occur. Evans (1989, 1995; Wason & Evans, 1975) has been most active in exploring this approach in a reasoning context.

Probabilistic theory

Unlike all the aforementioned theories, the probabilistic theory does not rely on logic for its normative model, but rather draws on probability theory (e.g., Bayesian probability theory). Reasoning is not about validity but is about maximising information gain to reduce uncertainty. A maximally informative statement is one that tells you something improbable or surprising, relative to your prior knowledge. People have the cognitive goal of reducing uncertainty by increasing informativeness, and they make conclusions designed to maximise informativeness. Oaksford and Chater (1994, 1995, 1996; Chater & Oaksford, 1999a, b) have developed this theory.

Stop Anxiety Attacks

Stop Anxiety Attacks

Here's How You Could End Anxiety and Panic Attacks For Good Prevent Anxiety in Your Golden Years Without Harmful Prescription Drugs. If You Give Me 15 minutes, I Will Show You a Breakthrough That Will Change The Way You Think About Anxiety and Panic Attacks Forever! If you are still suffering because your doctor can't help you, here's some great news...!

Get My Free Ebook


Post a comment