Thinking and Reasoning

Psychological Approaches

Jonathan St. B. T. Evans, Jonathan Evans

• Publisher: Taylor & Francis
• Year: 2013
• Print ISBN: 9781848723177
• eBook ISBN: 9781317820390
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

The subject of thinking is the oldest in the whole science of psychology, going back to well before the separation of the disciplines of philosophy and psychology. Originally published in 1983, this collection of up-to-date critical essays about thinking – with particular emphasis on reasoning – is written from the perspective of psychologists who are themselves actively engaged in research into the nature of human thought.

The editor's introduction identifies the major issues which have traditionally concerned students of human thought, and provides an historical background. It describes how at first the subject was studied by introspection, and how this method fell into disrepute at the end of last century. A satisfactory alternative has not yet emerged, although much recent work is based on the information-processing model, which sees the brain as a sophisticated computer. Consequently the papers presented in this volume deal with a wide range of issues, and a number of different experimental tasks and paradigms. They cover most current approaches to the theory and methodology of cognitive psychology, including problem solving, the relationship between language and thought, and reasoning.

1	The role of problem content in the selection task and in the THOG problem
	Richard A. Griggs

This chapter will be concerned with the role of problem content in two tasks developed by Peter Wason. One of these, the selection task (or four-card problem), is also the topic of WASON’s chapter in this volume. The second one, the THOG problem, is a more recent addition to the deductive reasoning research area and is not as well known as the selection task. In fact, the THOG problem was developed in part because of the prominence of the selection task (Wason, 1977a; 1981).

The selection task involves the logic of a conditional rule for its solution while the THOG problem involves the logic of exclusive disjunction. However, these tasks are not simply propositional logic problems. In addition to an understanding of these propositional logic rules, both tasks require the formulation and testing of hypotheses. Exactly how each task does this will become clear as they are introduced in more detail later.

Performance on the selection task, and to a lesser extent on the THOG problem because of its recent development, has figured prominently in arguments concerning human rationality (e.g. Cohen, 1981; and see WASON, this volume) and Piaget’s theory of formal operations (e.g. Wason, 1977b). The factor most crucial to these arguments is the effect of problem content; specifically, the claim of facilitation in performance for thematic, realistic content. This chapter will consider this claim by first examining closely the research concerning this facilitation for the selection task. The main questions to be addressed are (1) Is there really a thematic facilitation effect? and (2) If so, what is its origin and nature? Following the selection task discussion, the recent research on the role of content in the THOG problem will be reviewed. Because published work on the THOG problem is so scarce at present, other questions concerning performance on that problem will also be discussed.

The question of thematic facilitation has also been examined in the other deductive reasoning paradigms (e.g. categorical syllogism and propositional reasoning tasks). This research has been discussed very clearly and thoroughly by Evans (1982), and thus there is no need to review it again here.

Wason’s selection task

Wason’s selection task involves an implication rule usually in the ‘If P then Q’ form and an array of four cards representing P,

(not P), Q, and

(not Q). Symbolical material (e.g. letters of the alphabet and numbers) and affirmative antecedent and consequent are employed in the basic form of the task. For example, the subject might be presented with four cards showing E, K, 4, and 7, and the rule, ‘If there is an E on one side of a card, then there is a 4 on the other side.’ The subject would be instructed that each card has a letter on one side and a number on the other side and that the task is to select just those cards that are necessary to turn over in order to decide whether the rule is true or false.

The problem has been the subject of much research since its initial description in the literature (Wason, 1966; 1968), because it is extremely difficult for even highly intelligent subjects. Usually less than 10 per cent of the subjects make the correct selection (Manktelow and Evans, 1979). In an informal experiment reported by Dawes (1975) only one of five subjects with doctorates in mathematical psychology solved the task correctly! The correct answer is a selection of the P and the

cards (E and 7 in the example) since only a combination of P on one side and

on the other side can falsify the implication rule. In general, subjects make two prevalent errors: (1) although many choose the P-card, they fail to select the

-card, and (2) frequently select the Q-card instead. Wason and Johnson-Laird (1972) provide a review of the early literature on this task while Evans (1982) covers this and some of the more recent literature. (See WASON’s chapter in this volume for an informative, historical account of the task.)

Thematic-materials effects in the early 1970s

Beginning with Wason and Shapiro (1971/Experiment 2), several studies were published during the early 1970s claiming to demonstrate a thematic-materials effect – an improvement in performance brought about by the use of ‘thematic’ materials. Exactly what is meant by ‘thematic’ material is defined in each study. The words thematic (e.g. Wason and Shapiro, 1971), realistic (e.g. Johnson-Laird, Legrenzi and Sonino Legrenzi, 1972), and concrete (e.g. Gilhooly and Falconer, 1974) have all been used to describe such material. In general, ‘thematic’ material has been defined, to be anything except the arbitrarily related symbols and forms (e.g. letters of the alphabet, numbers, geometric patterns, etc.) that have been used in the basic form of the task. The term thematic will be employed in this chapter since it has evolved to be the most frequently used term to describe this effect. Only four of the studies examining the thematic-materials effect (Wason and Shapiro, 1971; Johnson-Laird et al., 1972; Bracewell and Hidi, 1974; and Van Duyne, 1974) reported substantial improvement (50 per cent or greater correct selection performance); thus, these will be described first.

Wason and Shapiro (1971/Experiment 2) used rules that were presented as claims by the experimenter about four journeys she had made on four different days of the week. The rules involved two cities (Manchester and Leeds) and two modes of transportation (train and car), such as ‘Every time I go to Manchester, I travel by car.’ All four possible permuted variants of the two cities and the two modes of transportation were used in rules in a between-subjects design. It should be noted that before the task was presented, subjects were given a familiarization deck of 16 cards which included the four actual selection-task cards (Manchester, Leeds, car and train). Subjects were instructed to state which cards would need to be turned over to determine the truth or falsity of the experimenter’s claim. Ten of 16 subjects (62–5 per cent) made the correct selection compared to two of 16 (12.5 per cent) in an abstract condition using the basic version of the task with letters and numbers.

Johnson-Laird et al. (1972) employed a different thematic scenario. They asked each subject to ‘imagine that he was a post-office worker sorting letters’ and to ‘select those envelopes that you definitely need to turn over to find out whether or not they violate the rule’ (p. 397). The two thematic rules used were ‘If a letter is sealed, then it has a 50 lire stamp on it’ and ‘A letter is sealed only if it has a 5d. stamp on it.’ Invariably the methodology of this experiment has been described inaccurately in secondary sources (e.g. Wason and Johnson-Laird, 1972). Griggs and Cox (1982) should be consulted for a discussion of the methodological problems in the Johnson-Laird et al. study. Briefly, they stem from a blank fifth envelope present in all the experimental arrays but not usually mentioned in secondary descriptions of the study. Regardless, the performance they observed for thematic material is the highest observed in a published study, 39/48 (81 per cent) correct selections.

Results comparable to those of Wason and Shapiro (1971) were observed by Bracewell and Hidi (1974). Bracewell and Hidi used towns and modes of transport like Wason and Shapiro, except they examined the contributions of two factors, natural relationship v. arbitrary relationship of the terms and concrete (thematic) v. abstract terms. The concrete material/natural relationship rules (e.g. ‘Every time I go to Ottawa, I travel by car’) and the abstract/arbitrary relationship rules (e.g. ‘Every time J is on one side, 2 is on the other side’) are of main interest here. In the concrete/natural relationship problems they observed 9/12 (75 per cent) correct selections and only 1/12 (8·33 per cent) correct responses in the abstract/arbitrary relationship problems. However, as pointed out by Griggs and Cox (1982) and Manktelow (1979), a procedural difference in the Bracewell and Hidi study may have contributed partially to this effect. Bracewell and Hidi told their subjects that ‘the conditional rule was not reversible.’ The overall effect of this instruction seemed to be a decrease in inclusion of the Q-card in subjects’ selections. The P and Q combination is always one of the most frequent selections for the basic version of the task. However, only 3/96 subjects in the Bracewell and Hidi study made this selection. In addition, when the order of the clauses in the implication rule was reversed (e.g. ‘I travel by car every time I go to Ottawa’), the thematic materials effect completely disappeared (only 2/12 subjects correctly responded in the concrete material/natural relationship condition). Thus, it appears that the instruction about non-reversibility of the conditional rule may have interacted with content and other task and subject factors to bring about the good performance in the concrete material/natural relationship condition. For further discussion of problems with the Bracewell and Hidi study, see Manktelow (1979).

Like Bracewell and Hidi (1974), Gilhooly and Falconer (1974) were interested in both the nature of the relationship between the terms and the nature of the terms in the selection task. They used the towns-and-modes-of-transport thematic material like Wason and Shapiro (1971) but observed a much weaker but significant effect (11/50 or 22 per cent correct selections in their concrete terms/concrete relation condition). However, the procedures for familiarizing the subjects with the task materials differed between the Gilhooly and Falconer study and Wason and Shapiro’s experiment. As indicated earlier, the familiarization deck in the Wason and Shapiro study included the four cards used in the actual selection task. Thus, subjects appear to have been exposed to the falsifying instance in each case. As Gilhooly and Falconer point out, although Wason (1968) found no effect of prior identification of a falsifying instance for an abstract form of the task and no effect was observed in the abstract condition of Wason and Shapiro (1971), exposure to the falsifying instance may have been a factor when thematic materials were involved. The relevant test card may more easily have evoked memories of the falsifying instance in the case of the more concrete material. Gilhooly and Falconer used a different familiarization technique so that this did not occur, and they observed a much weaker effect. It should be noted that Bracewell and Hidi (1974) used a familiarization deck that did not contain the problem cards and they observed substantial facilitation in their concrete material/natural relationship condition. However, as explained above, interpretation of their data is hindered by possible effects of their instructions about the non-reversibility of the rule.

Van Duyne’s (1974) thematic material consisted of student cards taken from some hypothetical register of students. On one side of a card was a student’s academic subject and on the other side, his university. The four thematic rules used were ‘Every student who studies physics is at Oxford,’ If a student studies philosophy then he is at Cambridge,’ ‘A student doesn’t study French, or he is at London,’ and ‘It isn’t the case that a student studies psychology and isn’t at Glasgow.’ The four cards, respectively, for each of these rules were Physics, Spanish, Oxford, and Cambridge; Philosophy, Physics, Cambridge, and Oxford; French, Law, London, and Bristol; and Psychology, Chemistry, Glasgow, and Edinburgh. Subjects were instructed to ‘select only those cards you definitely need to turn over in order to find out whether they violate the rule or not.’ Only the first two rules showed significant improvement relative to abstract counterpart conditions. On the conditional rule 14/24 (58 per cent) were correct, and on the universal rule 12/24 (50 per cent).

Van Duyne (1976) should also be mentioned because of some results from an unpublished study described in this paper and by Wason (1977b). In this unpublished study, Van Duyne not only varied the abstract-thematic dimension but also the arbitrariness of the thematic rules. Van Duyne’s thematic rules concerned postal matters. The arbitrary rule was ‘If there is L. B. MILL on one side of the envelope, then there is PRINTED PAPER REDUCED RATE on the other side.’ The non-arbitrary rule was ‘If there is PRINTED PAPER REDUCED RATE on one side of the envelope, then it must be left open.’ Van Duyne reports a high-level of performance (87 per cent and 98 per cent) in the two non-arbitrary, thematic conditions (white cards were used as stimulus materials in one non-arbitrary condition and real envelopes in the other) and a lower-level of correct selections (almost 50 per cent) in the arbitrary, thematic condition. The level of performance in the non-arbitrary conditions is even higher than that in the Johnson-Laird et al. study, except Van Duyne’s study was never subsequently published and only this brief description of it is available.

One final study from this period should be noted. Lunzer (1975) mentions that Lunzer, Harrison, and Davey (1972) found an ‘appreciable and highly significant’ effect due to thematic material (i.e. a relationship between the color of a lorry and whether or not it was full of coal or empty). However, Lunzer must be referring to the Lunzer et al. data for a reduced presentation array (Q, Q,

,

instead of the usual P,

, Q,

), because there is clearly no effect of content for complete arrays in the Lunzer et al. experiments.

The elusive thematic-materials effect in the late 1970s and early 1980s

Manktelow and Evans (1979) was the first in a series of recent failures to replicate the earlier studies finding thematic-materials effects. Manktelow and Evans conducted five experiments including an almost exact replication of the Wason and Shapiro study. The first four experiments examined rules involving food and drink, such as ‘If I eat haddock, then I drink gin.’ No evidence for a thematic-materials effect was found in any of the four experiments (Yachanin and Tweney, 1982; and Reich and Ruth, 1982/Experiment 1, have also failed to find an effect for this type of thematic material). In Manktelow and Evans’s fifth experiment, they also failed to replicate the Wason and Shapiro findings. The only methodological difference in the studies is that Manktelow and Evans showed the four days of the week for the journeys on four separate cards instead of on the four array cards as in the Wason and Shapiro study.

Except for one study (Pollard, 1981) in which a weak effect was observed, other recent attempted replications of the Wason and Shapiro study have also failed....

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Original Copyright Page
• Table of Contents
• Contributors
• Introduction
• 1 The role of problem content in the selection task and in the THOG problem
• 2 Realism and rationality in the selection task
• 3 The language and thought of disjunction
• 4 The role of ‘representativeness’ in statistical inference: a critical appraisal
• 5 Selective processes in reasoning
• 6 Thinking as a skill
• 7 Mental imagery in thinking and problem solving
• 8 Protocol analysis in problem solving
• Index