## Figure 163

Examples of the abstract and concrete (postal) versions of Wason's selection task, with an indication of how the different cards/envelopes are labelled for the classification of subjects' choices in experiments.

In logic, the rule in the selection task is a conditional if P then Q; P here is the statement that "there is a vowel on one side of the card" and Q is "there is an even number on the other side of the card". So, each of the cards can be re-expressed as follows: the E-card is P, the K-card is not-P, the 4-card is Q, and the 7-card is not-Q (see Figure 16.3). It is not a good idea to pick the Q card (i.e., 4-card) because when Q is true and the rule if P then Q is true, P can be either true or false (see the truth table in Table 16.1); irrespective of what is on the other side of the 4-card (a vowel or a consonant), the rule will be true. Thus, turning over this card will tell you very little logically (cf. Oaksford & Chater, 1994). Of course, many people who turn over the Q card may be making the fallacious affirmation of the consequent inference; they assume the rule If P then Q is true, they know Q to be true and so they can conclude that P must be on the other side (see Table 16.2). So, if not-P is on the other side they feel they can conclude that the rule is not true. The same sort of reasoning may account for turning over the K-card, except that in this case one is making an inference similar to the denial of the antecedent fallacy.

In contrast, the choices of the E-card (i.e., P) and the 7-card (i.e., not-Q) are correct because one is making logically valid inferences that may falsify the rule. When P is true and we turn over this card, then what we find on the other side will indicate whether the rule is true or false (see Table 16.1). If Q is on the other side then the rule is true, if not-Q then the rule is false (this is similar to a modus ponens inference). Similarly, if one turns over not-Q then a P on the other side will make the rule false and a not-P on the other side will make the rule true (see Table 16.2).

### Effects in the abstract .selection task

The difficulty of reasoning in this problem should be apparent. Typically, in abstract versions of the task (i.e., ones involving vowels and consonants) very few subjects make the correct choices. In Johnson-Laird and Wason's (1970) study only 5 subjects out of 128 chose the P and not-Q cards alone. The overwhelming majority of subjects choose either the P and Q cards (59 out of 128) or the P card alone (42 out of 128).

Originally, it was thought that subjects were trying to confirm rather than falsify the rule (see e.g., Wason & Johnson-Laird, 1972; and Chapter 15); they turned over the P-card to see if there is a Q (i.e., an even number) on the other side and the Q-card to see if there is a P (i.e., a vowel) on the other side. If they had wanted to falsify the rule they would have chosen the P-card to see if there is not-Q on the other side (i.e., a consonant) and not-Q to see if there is P (i.e., a vowel) on the other side of it. However, as we shall see, variants of the task involving more realistic materials lead to very different behaviour.

### Matching bias effects in the selection task

Evans has suggested that in such abstract versions of the task subjects manifest a non-logical, matching bias (Evans 1984 , 1998; Evans & Lynch, 1973; Wason & Evans, 1975). That is, subjects select those cards showing the symbols that are mentioned in the rule. So, when subjects are given another variant of the rule (i.e., "If there is a B on one side, there is not a 3 on the other side") they choose the B (P) and the 3 (not-Q) card because they are mentioned in the rule. In a separate conditional, truth-table task, Evans (1983) found that matching bias depended on the way the "negative cards" are presented; the negative cards, not-P and not-Q, can be presented as "explicit negatives" (e.g., not-P can presented as "not an A"), or as "implicit negatives" (e.g., not-P presented as "K"). Evans found that the use of explicit negatives reduced matching bias and facilitated subjects' performance on the task.

### Effects in thematic selection tasks

One of most researched effects on performance in the selection task is the change in subjects' performance when they are given "concrete" or "realistic" or "thematic" content in the task (see Bracewell & Hidi, 1974; Gilhooly & Falconer, 1974; Wason & Shapiro, 1971). Johnson-Laird, Legrenzi, and Sonino-Legrenzi (1972) used realistic materials; they asked subjects to imagine that they worked in a post office and had to detect violations in a rule, given letters of different types (in the pre-decimalisation days when unsealed letters were cheaper to send):

If a letter is sealed, then it has a 5d. stamp on it.

The envelopes provided were either sealed or unsealed and had a 4d. or a 5d. stamp on the side that was showing (see Figure 16.3). Again, subjects had to make just those choices needed to determine if the rule had been violated. Johnson-Laird et al. also used an abstract version of the task involving an abstract rule (i.e., "If there is a D is on one side, then there is a 5 on the other side"). They found that, of the 24 subjects who attempted the tasks, 92% (22) produced the correct choices on the realistic version and only 8% (2) were successful on the abstract version.