LANGUAGE, THINKING AND MEMORY

1
LANGUAGE, THINKING AND MEMORY

• HYPOTHESIS TESTING

2
WASON'S SELECTION TASK

• Four cards, each with a letter on one side and a
  number on the other - only one side showing
• E K 4 7
• P not-P Q not-Q
• Which cards must be turned over to see if the
  rule If a card has a vowel on one side, it has
  an even number on the other applies to all the
  cards?
• ANSWER The P card (E) and the not-Q card (7)

3
POSSIBLE STATUS OF CARDS

• Possible types of letters and numbers on the
  reverse of cards in the selection task, and their
  implications for the status of the cards
• card (type) number or status of card
• letter on reverse
• --------------------------------------------------
  ---------------------------------------------
• P (E) even number confirming
• odd number falsifying
• Not-P (K) even number irrelevant
• odd number irrelevant
• Q (4) vowel confirming consonant irrelev
  ant
• Not-Q (7) vowel falsifying
• consonant irrelevant

4
MATCHING BIAS (Evans)

• Experiment using Four Rules
• If there is a vowel on one side of the card,
  there is an even number on the other side.
• If there is a vowel on one side of the card,
  there is not an even number on the other side.
• If there is not a vowel on one side of the card,
  there is an even number on the other side.
• If there is not a vowel on one side of the card,
  there is not an even number on the other side.

5
MATCHING BIAS - LOGICAL TYPES OF CARDS

• P not Q not
• P Q
• --------------------------------------------------
  --
• Rule 1 E K 4 7
• Rule 2 E K 7 4
• Rule 3 K E 4 7
• Rule 4 K E 7 4
• The asterisked cards (E and 4) are those
  explicitly mentioned in the rule.

6
CONCRETE CONTENT VERSION OF THE SELECTION TASK

• Manchester Leeds Car Train
• P not-P Q not-Q
• Rule Every time I go to Manchester I go by car.

7
DEONTIC RULES VERSIONS OF THE SELECTION TASK

• If an envelope is sealed, it must have a 5d
  stamp.
• If someone is to drink alcohol (in this bar),
  they must be over 18.
• If someone enters the country, they must have a
  cholera vaccination certificate.
• If a man eats cassava root, then he must have a
  tattoo on his face.
• If you clean up split blood, you must wear rubber
  gloves.
• If you tidy your room, then you may go out to
  play (Child's perspective vs parent's perspective)

8
THEORIES OF ENHANCED SELECTION TASK PERFORMANCE

• Memory Cueing (Griggs and Cox) - stamps, drinking
  laws
• Pragmatic Reasoning Schemas (Cheng and Holyoak) -
  cholera
• Reasoning about costs and benefits in Social
  Contracts (Cosmides) -cassava root
• Search for violators of rules (Gigerenzer and
  Hug) - cassava root etc.
• Practical Reasoning (Manktelow and Over) - room
  tidying
• Making implicit mental models explicit
  (Johnson-Laird Byrne)

9
THE THOG PROBLEM

• Four shapes
• black square
• black circle
• white square
• white circle
• Experimenter I am thinking of one colour (black
  or white) and one shape (square or circle). Any
  figure that has either the colour I am thinking
  of, or the shape I am thinking of, but not both,
  is a THOG.
• Given that the black square is a THOG what, if
  anything, can you say about whether the other
  figures are THOGs?

10
THOG PROBLEM - SOLUTION

• black black white white
• Rule square circle square circle
• --------------------------------------------------
  -----------------------------------------
• black or THOG non-THOG non-THOG THOG
• circle (black) (both) (neither)
  (circle)
• white or THOG non-THOG non-THOG THOG
• square (square) (neither) (both)
  (white)

11
THOG - THE INTUITIVE ERROR

• The Intuitive Error the white circle is not a
  THOG and the other two figures are THOGs (or
  their status is indeterminate).
• Intuitive because white circle and black square
  have no common features white square and black
  circle,share a feature with the known THOG.

12
THE INTUITIVE ERROR (cont)

• More specifically
• Some subjects confuse the two properties of the
  exemplar THOG (black and square) with the
  properties that the experimenter has in mind.
• Most subjects fail to construct the two possible
  definitions of a THOG.
• Newstead and Griggs both difficulties arises
  from a failure of to distinguish data (the
  figures) from hypotheses (about what properties
  experimenter has in mind and about what defines a
  THOG).
• Confusion is easy because each hypothesis (e.g.
  black and circle) can be interpreted as a
  description of a figure (e.g. the black circle).

13
TWO STUDIES WITH IMPROVED PERFORMANCE IN THE
THOG TASK

• 1. Newstead and Griggs PUB problem
• A person gives cards to each of four friends in a
  pub and keeps one for himself. He says he will
  buy a dinner for those friends who have a card
  with a same shape or same colour as his, but not
  both. Then he says, by this rule I must buy a
  dinner for John, can you work out who else I must
  buy one for?
• Main character's card hypothesis
• four friends' cards data

14
TWO STUDIES WITH IMPROVED PERFORMANCE IN THE THOG
TASK

• 2. Girotto and Legrenzi SARS
• Standard version of the problem, but gave another
  nonsense name (SARS) to the card corresponding to
  the colour and shape the experimenter has in
  mind.

15
THE 2-4-6 PROBLEM

• Subject has to discover a rule governing
  sequences of three numbers.
• The sequence 2-4-6 obeys the rule.
• The subject asks the experimenter whether other
  sequences obey the rule.
• The experimenter answers yes or no.
• Actual rule any three numbers in ascending order.

16
THE 2-4-6 PROBLEM -RESULTS

• Subjects tend to ask about positive examples of
  the rule they have in mind, though asking about
  negative examples would be more informative.
• Wason calls this Confirmation Bias
• BUT it may be just a Positive Test Strategy
  (Klayman and Ha).
• This strategy is useful when the real rule is
  specific (cf. the 2-4-6 task, in which it is
  rigged to be very general).

17
THE 2-4-6 PROBLEM - MORE RESULTS

• DAX-MED version (Gorman et al.) the sequences of
  numbers are classified as DAX (those that obey
  the rule) and MED (those that don't).
• Performance is improved, perhaps because of a
  better representation of the problem.

18
ARTIFICIAL SCIENCE STUDIES

• Mynatt, Doherty, and Tweney (1977, 1978)
• Computer-simulated miniworld in which particles
  could be fired at figures, some protected by
• invisible boundaries. Subjects had to discover
  what influenced motion of particles.
• Mynatt et al. argue that people understand the
  logic of falsification, but fail to eliminate
  hypotheses that are entirely misguided (e.g.
  shape of objects, rather than brightness).

19
ARTIFICIAL SCIENCE STUDIES (cont.)

• In the second study, with a more complex set of
  rules, disconfirmatory instructions led to poorer
  performance. Mynatt et al. suggest the need to
  establish, in the early stages of an
  investigation, at least one viable hypothesis
  that can account for a range of data.