Online research in philosophy

We examine the role of working memory's central executive in the mental model explanation of propositional reasoning by using two working memory measures: the classical “reading span” test by Daneman and Carpenter (1980) and a new measure. This new “reasoning span” measure requires individuals to solve very simple anaphora problems, and store and remember the word solution in a growing series of inferential problems. We present one experiment in which we check the involvement of the central executive in conditional and disjunctive inference tasks and compare predictions of the new reasoning span test with those of the classical reading span test. The results of the experiment confirm that reasoning responses, which according to mental model theory require high cognitive work, are predicted by working memory measures. Results also show that some reasoning responses are probably obtained by means of superficial biases or strategies that do not load working memory. The reasoning span test, which involves the central executive to a greater degree, predicts reasoning performance better than the reading span test. The significance and possibilities of the new measure in studying reasoning are discussed.

This paper reports three studies of temporal reasoning. A problem of the following sort, where the letters denote common everyday events: A happens before B. C happens before B. D happens while B. E happens while C. What is the relation between D and EEfficacylls for at least two alternative models to be constructed in order to give the right answer for the right reason (D happens after E). However, the first premise is irrelevant to this answer, and so if reasoners were to ignore it, then they would need to construct only one model. Experiment 1 showed that one-model problems were answered faster and more accurately than multiple-model problems. When the question preceded the premises in the statement of the multiple-model problems there was a slight tendency for the latencies of response to speed up in the predicted way. Experiment 2 modified the procedure, in part by using practice problems with many irrelevant premises, so that reasoners might grasp the advantage of ignoring them. Its results showed that when the premises preceded the question, the multiple-model problems were significantly harder than one-model problems. But when the question was presented first, the difference was significantly reduced in line with the theory's prediction. Experiment 3 used only problems with valid conclusions (i.e., one-model problems and multiple-model problems), and so the construction of multiple models was never necessary. However, there was still a significant difference between one-model problems and multiple-model problems.

This paper reports four experiments investigating whether model construction of linear reasoning problems is open to strategic decisions. A reversed choice/nochoice paradigm was used in which reasoners first had to apply two model construction strategies (acronym and rehearsal strategy) to two problem sets. Next, they could choose freely among the two strategies to apply to a new problem set. Experiment 1 showed that reasoners selected the strategy that they experienced as the most accurate one in the no-choice phase. Moreover, in Experiment 2, it was found that reasoners adapted their strategy choice to changing problem features, to use the most suitable strategy for premise encoding. Experiments 3 and 4 generalised these findings to more complex linear reasoning problems with a mixed sentence frame and a semi-continuous presentation of the premises, and to two-model problems. On the basis of these results, we argue that strategic decisions influence model construction in linear reasoning.

We report three experimental studies of reasoning with double conditionals, i.e. problems based on premises of the form: If A then B. If B then C. where A, B, and C, describe everyday events. We manipulated both the logical structure of the problems, using all four possible arrangements (or ''figures" of their constituents, A, B, and C, and the believability of the two salient conditional conclusions that might follow from them, i.e. If A then C , or If C then A . The experiments showed that with figures for which there was a valid conclusion, the participants more often, and more rapidly, drew the valid conclusion when it was believable than when it was unbelievable. With figures for which there were no valid conclusions, the participants tended to draw whichever of the two conclusions was believable. These results were predicted by the theory that reasoning depends on constructing mental models of the premises.

Johnson-Laird (1983) has argued that spatial reasoning is based on the construction and manipulation of mental models in memory. The present article addresses the question of whether reasoning about time relations is constrained by the same factors as reasoning about spatial relations. An experiment is reported that explored the similarities and the differences in the performance of subjects in comparable spatial and temporal reasoning tasks. The results indicated that, in both the temporal and the spatial content domains, the data were in agreement with the view that subjects solved problems by constructing models in memory rather than with a logical rule conceptualisation of reasoning. An analysis of the premise-reading times on the basis of premise-linking order provided support for an on-line process of mental models construction, and offered an explanation for the finding that spatial problems that did require an inference of transitivity were easier than problems that did not. No essential differences in processing and performance were observed across the two content domains, although in the time domain the correctness data were in agreement with both the mental models theory and the logical rules view. The results are discussed with respect to the mental models theory and the structural characteristics of the problems.

Four experiments investigated uncertainty about a premise in a deductive argument as a function of the expertise of the speaker and of the conversational context. The procedure mimicked everyday reasoning in that participants were not told that the premises were to be treated as certain. The results showed that the perceived likelihood of a conclusion was greater when the major or the minor premise was uttered by an expert rather than a novice (Experiment 1). The results also showed that uncertainty about the conclusion was higher when the major premise was uttered by a novice and an alternative premise by an expert, compared to when the major premise was uttered by an expert and the alternative by a novice (Experiment 2). Similarly, the believability of a conclusion was considerably lower when the minor premise was uttered by a novice and denied by an expert, as opposed to when an expert uttered the minor premise and a novice denied it (Experiment 3). Experiment 4 showed that the nature of the uncertainty induced by a denial of the minor premise depended on whether or not the context was a conversation. These results pose difficult problems for current theories of reasoning, as current theories are based on the results of experiments in which the premises are treated as certain. Our discussion of the results emphasises the importance of pragmatics in reasoning, namely, the role of general knowledge about the world in assessing the probability of a premise uttered by an expert or a novice and the role of interpretations of the premise based on pragmatic inferences in revising these initial probabilities.

Two experiments investigated the mental representation of spatial and nonspatial two-dimensional problems. The experiments were designed to contrast opposite predictions of the model theory of reasoning and the formal rules of inference theories. Half of the problems required more inferential steps but only one model, whereas the other half required fewer inferential steps but two models. According to the inference rules, theory problems that require more inferential steps should be harder, whereas the model-based theory predicts that problems that require two models should be harder. In Experiments 1a and 1b we measured the problem solving time and the percentage of errors. In Experiments 2a and 2b the problems were presented segmented in two different displays. We measured the comprehension time for each display, the question answering times, and the percentage of errors. The results of all experiments supported the model theory predictions in both spatial and nonspatial domains.

Relatively little is known about those who consistently produce the valid response to Modus Tollens (MT) problems. In two studies, people who responded correctly to MT problems indicated how “convinced” they were by proofs of conditional reasoning conclusions. The first experiment showed that MT competent reasoners found accurate proofs of MT reasoning more convincing than similar “proofs” of invalid reasoning. Similarly, there was a tendency for MT competent reasoners to find an initial counterfactual supposition more convincing than did people who were less competent in MT. The second experiment showed that when individuals produced the correct MT response, and found correct MT proofs to be more convincing than “bogus” proofs, they were also less likely to find the conclusions to Denying the Antecedent, or Affirming the Consequent problems valid, compared to individuals who could not discriminate between valid and bogus MT proofs. These findings are discussed in terms of both their implications for the mental logic and mental models positions, and individual differences in System 1 and System 2 reasoning.

The aim of the present research was to develop a difficulty model for logical reasoning problems involving complex ordered arrays used in the Graduate Record Examination. The approach used involved breaking down the problems into their basic cognitive elements such as the complexity of the rules used, the number of mental models required to represent the problem, and question type. Weightings for these different elements were derived from two experimental studies and from the reasoning literature. Based on these weights, difficulty models were developed which were then tested against new data. The models had excellent predictive validity and showed the relative influence of rule based factors and factors relating to the number of underlying models. Different difficulty models were needed for different question types, suggesting that people used a variety of approaches and, at a wider level, that both mental models and mental rules may be used in reasoning.

The aim of this paper is to investigate two related aspects of human reasoning, and use the results to construct an automated theorem prover for the predicate calculus that at least approximately models human reasoning. The result is a non-resolution theorem prover that does not use Skolemization. It involves two central ideas. One is the interest constraints that are of central importance in guiding human reasoning. The other is the notion of suppositional reasoning, wherein one makes a supposition, draws inferences that depend upon that supposition, and then infers a conclusion that does not depend upon it. Suppositional reasoning is involved in the use of conditionals and reductio ad absurdum, and is central to human reasoning with quantifiers. The resulting theorem prover turns out to be surprisingly efficient, beating most resolution theorem provers on some hard problems.