Tolman
Tolman
Tolman
Tolman
(1886-1959)
Chapter 12
Edward C. Tolman
1. Born (1886) in West Newton, Massachusetts. 2. B.S from MIT. PhD from Harvard. 3. Studied under Koffka. 4. 1915-1918 taught at Northwestern University. Released from university. Pacifism!
www.uned.es
(1886-1959)
2
Edward C. Tolman
5. Moved to University of California-Berkeley and remained till retirement. 6. Dismissed from his position for not signing the loyalty oath. Fought for academic freedom and reinstated. 7. Quaker background therefore hated war. Rebel in life and psychology.
www.uned.es
(1886-1959)
3
Edward C. Tolman
8. Did not believe in the unit of behavior pursued by Pavlov, Guthrie, Skinner and Hull. Twitchism vs. molar behavior. 9. Learning theory a blend of Gestalt psychology and behaviorism. 10. Died 19 Nov. 1959.
www.uned.es
(1886-1959)
4
Comparison of Schools
Behaviorism Behaviorists believed in elements of S-R associations. Observation and Experimentation Approach: Behavioral Gestalt
Gestalt psychologists believed in the whole mind or mental processes. Observation, Experimentation and Introspection
Approach: Cognitive
5
Differences
Behaviorism Pavlov, Watson, Thorndike, Guthrie, Skinner and Hull proposed S-R associations meant learning. Study molecular behaviors. Dont explain behavior. Purposive Behaviorism
Tolman suggested, S-R associations were not beneficial for analyzing learning. Instead analysis of S-S associations were important in understanding learning.
Study molar behaviors. Behavior is goal-directed.
8
Similarities
Behaviorism Purposive Behaviorism Both agree that study of animal behavior is important not the study of animal mind.
Latent Learning
To show rats can learn by exposure (latent learning), Blodgett (1929) took three groups of rats and allowed them to explore 6 unit alley T-maze. One group was not reinforced at all, the other two were reinforced after 3 and 7 days.
10
Average Errors
11
Days
Latent Learning
Tolman and Honzik (1930) in a 14 unit alley T-maze, replicated Blodgett (1929) study and concluded that rats learnt latently while exploring the maze.
12
Average Errors
8 6 4 2 0 1 2 3 4 5 6 7 8 9 10
11 12 13 14 15
16 17
Days
1. Reinforced group (green) shows steady improvement in performance. 2. For the reinforcement day 11 group (red) sudden improvement in performance is observed after 11th day.
13
No Reinforcement Group
3. A slight improvement in performance of no reinforcement group (blue) was also observed. 4. According to behaviorists (chiefly Hull), rats were getting reinforcement because they were removed from the maze after reaching the goal box. 5. Tolman and Spence disagreed, they said that this slight improvement in performance was starkly miniscule to the enormous improvement in the reinforcement day 11 group (red).
14
Experimental Apparatuses
Experiments with VTE involved classical and more modern apparatuses, used in visual discrimination procedures (Lashley, 1930; Tolman, 1948).
17
Visual Discrimination
1. To study VTE and associated run, Tolman and his colleagues trained rats to visually discriminate among three sets of stimuli. Between a white and a black door (easy), between a white and a medium grey door (harder), and between a white and a light grey door (hardest).
18
Visual Discrimination
2. For the easy task (White-Black) rats took fewer days to learn the discrimination than harder (White-Medium Grey) or the hardest (WhiteLight Grey) task.
6 5
Correct Runs
4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
White-Black Discrimination White-Medium Grey Discrimination White-Light Grey Discrimination
Days
19
Average VTEs
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Days
20
VTEs: Conclusions
1. Rats were able to select and compare stimuli early on for the easy task; made more VTEs, and correct runs. For harder tasks they took more trials before they could make more VTEs and more accurate runs. It takes rats many more trials to make fine discriminations between white and grey doors. 2. Accurate discriminations (VTEs) and effective runs further support the idea of building cognitive maps.
21
Response Learning
1. Conventional behaviorists emphasize S-R learning, i.e., animals learn a specific response(s). Tolman and colleagues showed that response learning was difficult and took longer time for the animals to acquire.
22
Place Learning
2. Tolman and colleagues clearly showed place learning was much easier to learn. Rats made fewer errors and learnt the task in fewer trials. Cognitive maps were easily formed and rats rerouted their paths to reach the goal.
23
Results
Data suggested that rats that learnt a place made fewer errors and reached zero errors on the 9th trial, whereas response learners continued to make errors to the end.
8 7 6 5 4 3 2 1 0 -1
Total Errors/Group (N = 8)
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33
Trials
24
Place Learning
In the training period, rats learned the maze from A to G. In the testing phase, the maze was changed to a sun-burst pattern, alley C-D was blocked and a choice was given to select alleys 1-12.
25
Results
Since alley 6 spatially matched up with the goal box of the maze in the testing period. This alley was maximally selected by a large percentage of rats.
40 35
34.0
Percentage of Rats
30 25 20 15 10 5 0 12 11 10 9 8 7 6 5 4 3 2 1
2.0 4.0 7.5 2.0 7.5 6.0 8.0 4.0 4.0 2.0 17.0
Pathway
26
Cognitive Map
1. For Tolman, place learning lead to development of comprehensive cognitive map of the maze. 2. Cognitive map is a selective mental map, a broad picture of the environment, which develops by environmental exploration. 3. Once the lay of the land (cognitive map) is developed, the animal can reach its goal from any direction, individual responses, even individual routes to goal do not matter.
27
Hypotheses
Krech and Tolman (1932) used a complicated maze to study how rats used strategies (hypotheses) to reach the goal box. They changed the colors of the doors (dark, light), making them swing or freeze etc.
28
30
10
Latent Extinction
6. Latent extinction refers to decrease in responses which occur as a result of non-reinforced experience in the goal box. Rats trained in a maze with reinforcement in goal box, were then placed directly into an empty goal box by the experimenter. Rats when reintroduced into the maze showed hesitant responses to go to the goal box for they did not expect the reinforcement.
31
Reinforcement Expectancy
1. For Tolman, changing reinforcement meant change in expectations and thus performance. Switching from one reward to another changes expectations which can favorably or adversely affect performance (Elliot, 1932). 2. For S-R theorists changing comparable reinforcements would not bring a change in behavior. Learning would be equivalent across reinforcing conditions.
32
Results
Rats in the bran mush group did consistently better than the sunflower seed group. When diet for bran mush group changed, their expectations changed, and thus their performance.
10 9 8 7
Average Errors
6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
33
DailyTrials
11
Conclusions
1. Hullians would argue that the incentive (K) value of bran mush was higher than sunflower seeds, so when bran mush group was given sunflower seeds to eat, their performance (SER) decreased. 2. Tolman argued that after diet change, bran mush group performed poorly to sunflower seed group, and continued to do so, even when sunflower seed groups performance improved. So when expectations were not met, bran mush group performed poorly.
34
35
36
12
Behavior Analysis
If an animal has been previously trained to turn left in a T-maze, his tendency to turn left will be given by the formula given below: BL Behavior Turning Left = B + B L R
=
6 6 + 4
= 60%
37
Accumulated Knowledge
The formula OBO represents animals all accumulated knowledge that comes from making B L and BR responses. Animal learns what leads to what.
BL BR
O C:
BL: OL:
OC
Choice point
Behavior turning left Complex stimuli on left
BL BL + BR
38
Environmental Variables
OBO in essence represents learning trials manipulated by the experimenter, thus an independent variable. However this is not the only variable that affects the animals behavior.
Independent Variable Maintenance schedule Appropriateness of goal object Types or modes of stimuli Motor responses required. Patterns of preceding and succeeding mazes Symbol M G S R P
39
13
Individual Variables
In addition to independent variables the animal brings its individual difference variables to the learning situation.
Independent Variable Heredity Age Previous training Endocrine function, drugs etc. Symbol H A T E
40
Intervening Variables
Tolman (and later Hull) used operationally defined intervening variables to functionally connect independent and dependent variables.
Independent Variable
Maintenance schedule Appropriateness of goal object Types or modes of stimuli Motor responses required Number of trials
Intervening Variable
Demand Appetite Differentiation Motor Skill Hypotheses
41
Relationship of Variables
Independent Variables M G S R OBO P
f2
H.A.T.E H.A.T.E H.A.T.E H.A.T.E H.A.T.E H.A.T.E
f3
Dependent Variable
BL BL + BR
42
14
Theory of Learning
1. MacCorquodale and Meehl (1953) attempted to formulate Tolmans theory into precise terms and concepts so that they could be tested. Briefly they used S1 R1 S2 to formulate and summarize Tolman's learning theory. 2. The formula simply says what leads to what, e.g., seeing a friend saying hello seeing your friend smile. 3. Tolman took a lively approach toward his theory and despite its weaknesses, he said that is the way he understood learning and psychology. 43
Kinds of Learning
In the end Tolman concluded that there are many kinds (6) of learning, each with its own particular characteristics. 1. 2. 3. 4. 5. 6. Cathexes Equivalence Beliefs Field Expectancies Field Cognition Modes Drive Discriminations Motor Patterns
44
Cathexes
1. Cathexis (pl. cathexes) is a learnt tendency to associate certain objects with certain drive states. Fish eating people tend to satisfy their hunger with fish, others perhaps with grains, etc. 2. When an individual is positively cathected (hungry) he approaches desired food (say, fish), if he is negatively cathected (hungry) he avoids an undesired food (say, worms). 3. No difference between Tolman and S-R theorists on this learning. 45
15
Equivalence Beliefs
When a subgoal has the same affect as the goal itself it is called equivalence belief. If getting a high grade in class temporarily reduces a students need for love, we say that grade have served as equivalence belief. This is equivalent to secondary reinforcement in S-R theories.
46
Field Expectancies
Field expectancies are developed in the same way as a cognitive map. A form of learning in which the organism learns what leads to what. This is S-S learning, unlike S-R learning. Reinforcement in this form of learning is confirmation of a hypothesis.
47
48
16
Drive Discriminations
Learning based on discriminating drives. Animals could be trained to take one route in two-way maze when they were hungry and another when they were thirsty (Hull, 1933; Leeper, 1935). If the drives are not clear, goals will not be clear and thus inappropriate behavior will result.
49
Motor Patterns
Motor patterns are a form of learning that is based on an association of stimuli and movements. This learning resembles Guthries learning.
50
Evaluation
Contributions
Tolman incorporated Gestalt ideas into behaviorism, which matches current ideas in cognitive psychology. Certainly not regressive. Learning without reinforcement (latent learning) hallmarks new theoretical approach.
Criticisms
Regression to mentalistic psychology, no practical applications (Malone, 1991).
Theory not easily testable. Large number of independent, dependent and intervening variables
51
17
Questions
23. How would Tolman explain to behaviorists what does a rat do when traverses a maze and learns it? 24. Describe latent learning and vicarious trial and error (VTE) response? What evidence did Tolman find for latent learning? 25. Differentiate between response and place learning. Relate theoretical (behavioral, cognitive) underpinnings to explain these differences.
52
18