Chapter 11 Spatial, Motor-Skill, & Implicit Learning

1. PSY 445: Learning & Memory Chapter 11 Spatial, Motor-Skill, & Implicit Learning

2. Procedural Knowledge The ability to quickly perform various cognitive, perceptual, and motor operations • Helps us to achieve skilled behavior; appears mostly implicit Three types will be covered in this chapter: Spatial Learning • Knowing how to get from place to place in the environment Motor-skill Learning • Knowing how to perform coordinated bodily movements quickly and accurately Implicit Learning • Knowing the underlying rules that govern complex sequences of behavior

3. Route vs. Survey Maps Route Maps • The knowledge of a series of routes, directions, or paths through a spatial environment • Characterized by knowledge of sequential locations Survey Knowledge • An abstract representation of the environment, placing specific routes in context with the surrounding area • Ariel overview • Cognitive map

4. The great Debate: Behaviorists vs. Cognitivists • Cognitivists • Learning takes place in the mind, not in behavior • The formation of mental representations of the elements of a task and the discovery of how these elements are related • Forming a cognitive map of the environment (Tolman, 1948) • Learning the correct place • Behaviorists • Learning involves the formation of associations between specific actions and specific events (stimuli) in the environment (Hull, 1949) • Learning the correct response Place vs. Response Studies

5. Place vs. Response Studies Series of studies with inconsistent results (results appear to be influenced by who the researchers were) • In one experiment, after rats received the reward, the researchers simply rotated a maze • Tolman’s theory (place) would predict that the rats would check their cognitive map for the location of the maze within the room, and make a left turn to compensate for the change in starting position • Hull’s theory (response) would predict rats would make a right turn as they have been trained to do See next slide for results 

6. Place vs. Response Studies Tolman, Ritchie, & Kalish (1946, 1947) • Rats made left turn • Used maze with flat alleys, no walls, and elevated • Distinctive features in the room were clearly seen from the maze; encouraged cognitive learning Hull (1949) • Rats made right turn • Maze alleys were enclosed by walls • Maze itself was surrounded by curtains • No prominent cues for the rat to orient itself within the room; encouraged response learning Edward Tolman (1886-1959) Clark Hull (1884-1952)

7. Tolman, Ritchie, & Kalish (1946) Start 2 This maze had no walls or roof so that rats could see “landmarks” in the room such as a window, door, or lamp. Place vs. Response Studies On a random half of the trials, the rats started from Start Box 1, and on the other half they started from Start Box 2. Goal 2 Goal 1 Start 1 GROUP P always found food in Goal Box 1. GROUP R found food in Goal Box 1 when they started from Start Box 1 but received food in Goal Box 2 when they started from Start Box 2.

8. Start 2 Cognitive theory predicted that GROUP P would learn faster because they only had to learn one cognitive map. Place vs. Response Studies Behavior theory predicted GROUP R would learn faster because they only had to learn one sequence of movements at the choice point—a right turn. Goal 2 Goal 1 Start 1 GROUP P always found food in Goal Box 1. GROUP R found food in Goal Box 1 when they started from Start Box 1 but received food in Goal Box 2 when they started from Start Box 2. Tolman, Ritchie, & Kalish (1946)

9. Start 2 What’s YOUR prediction? Place vs. Response Studies Are you a behaviorist or a cognitivist? GROUP R GROUP P Goal 2 Goal 1 Start 1 GROUP P always found food in Goal Box 1. GROUP R found food in Goal Box 1 when they started from Start Box 1 but received food in Goal Box 2 when they started from Start Box 2. Tolman, Ritchie, & Kalish (1946)

10. Start 2 What’s YOUR prediction? Place vs. Response Studies Are you a behaviorist or a cognitivist? GROUP R GROUP P Group P learned faster. But Goal 2 Goal 1 Later studies found that if the maze had a roof so the rats couldn’t see things in the room, response learning was faster. Start 1 GROUP P always found food in Goal Box 1. GROUP R found food in Goal Box 1 when they started from Start Box 1 but received food in Goal Box 2 when they started from Start Box 2. Tolman, Ritchie, & Kalish (1946)

11. Start 2 What’s YOUR prediction? Place vs. Response Studies Are you a behaviorist or a cognitivist? GROUP R GROUP P Group P learned faster. Goal 2 Goal 1 Both response and place learning occur. Which type is faster depends on what cues are available. So both the cognitive and behavioral views turned out to be right! Start 1 GROUP P always found food in Goal Box 1. GROUP R found food in Goal Box 1 when they started from Start Box 1 but received food in Goal Box 2 when they started from Start Box 2. Tolman, Ritchie, & Kalish (1946)

12. Place vs. Response Studies Interpretation • The place vs. response controversy taught us that either specific responses or cognitive maps may be learned: rats and people are flexible in their use of whatever cues are available

13. The Radial Maze Olton & Samuelson (1976) Procedure • 8 arms • All baited • Rat visits arms until all food is found • Number of visits is behavioral measure • 8 is minimum • Pattern of visits is also recorded Radial Maze 

14. The Radial Maze Task requires procedural memory • Rat must learn rules of the game: layout of maze, return trips to visited arms should be avoided, and so on Task also requires working memory • Rat must remember where it has been in order not to repeat a visit • At end of trial, rat can erase working memory and retain procedural memory

15. The Radial Maze Results • Rats do very well in this task, achieving an accuracy level of 7.6 different maze arms among the first 8 choices after only 15 trials Interpretation • Cognitive-mapping is convincingly demonstrated Click on picture  Olton & Samuelson (1976) After 7 days of training

16. The Radial Maze Alternative Explanations • Rats just enter arms in sequence thus assuring themselves of getting food and easing WM requirements • No - rats do not visit same arms in same order every day - pattern of arm visits is nearly random • Perhaps rats can smell food at end of arms or smell scents in visited arms • No - these possibilities have also been eliminated as dousing maze with after-shave lotion does not impair performance • Also, if after rat has made several choices, arms that it has chosen are again baited with food, then rat does not return to those arms

17. The Radial Maze Extramaze Cues • Rats seem to be tuned in to these cues which are outside the maze • Apparently, this allows them to keep track of entered and unentered maze arms • If one rotates maze so that spatial cues outside maze no longer give accurate information about where rat has and has not been, then rat’s performance deteriorates • Even though odor cues are available, rat makes mistakes by visiting locations that used to contain unvisited arms, but now, after maze rotation, contain arms that were already visited • It seems as if rat masters task by learning the maze perhaps by constructing a cognitive map in procedural memory • Rat then uses its working memory to keep track of where it has already been

18. Morris (1981) Morris Water Maze Procedure • A rat is placed in a small swimming pool in which the water is clouded by the addition of powdered milk • A hidden platform is located just under the water; rats’ goal is to learn the location of this platform • From trial to trial, the starting location of the pool is varied • Animal must learn the location of the platform on the basis of cues of the room Click on picture  Morris Water Maze

19. Morris Water Maze Results • On first trials, animals spend much time searching for the platform • Over trials, the animals become faster and follow more direct routes to get to the platform Interpretation • Rat appears to be encoding spatial relationships Morris (1981)

20. Maze Learning & The Brain SPATIAL LEARNING – ROLE OF THE HIPPOCAMPUS • Hippocampus

21. Maze Learning & The Brain • Hippocampal lesions – humans, other mammals • not all memories lost • memories of facts or events (explicit, declarative, episodic) impaired or lost • post-lesion memory impaired only • pre-lesion memory intact • procedural, implicit memories ok • important for acquisition & memory of types of new information • spatial information in particular LIMBIC SYSTEM 

22. SPATIAL LEARNING – ROLE OF THE HIPPOCAMPUS • Effects of hippocampal lesions – rats, radial arm maze • cued learning (informed that arm visited) ok • spatial impaired* • perseveration *

23. SPATIAL LEARNING – ROLE OF THE HIPPOCAMPUS • Effects of hippocampal lesions – rats, water maze • spatial impairment  longer circuitous routes • swimming impaired or just enjoy a good swim? • 3-part experiment  spatial task  impaired ability to find platform  cued task  can swim, would rather not  spatial task  still impaired... no memory!

24. Triple Dissociation of Limbic Area McDonald & White (1993) Hippocampal lesions caused impaired learning in regular radial –maze task • They consistently enter already visited arms Lesions of amygdala impaired association learning • They could not figure out that only lighted maze arms had food Lesions of striatum impaired learning of reinforcement • They did not repeat choices of reinforced arms In each case, the other two forms of learning were not affected

25. Landmarks There are certain elements of the environment that by virtue of their distinctive features (size or shape) or their meaning (historical or social) stand out from other features of the environment • The extramaze cues used by rats in place learning or the radial maze – doors, windows, light fixtures, etc. • Landmark recognition does not seem to fade with age as is the case with spatial memory overall External Landmarks • readily perceived • importance of learning the environment • direct sensory input (visual observation)

26. Schemas in Spatial Memory Spatial knowledge can be organized hierarchically in schemas • Spatial schemas have two prominent effects on memory: • Distortion in cognitive maps • Organization in spatial memory

27. Distortion in Cognitive Maps Spatial schemas distort recall due to the averaging, normalizing, or rounding off that occurs when a generalized map is acquired Stevens & Coupe (1978) • State locations are often used (incorrectly) to infer the relative locations of cities Spatial schemas seem to have a preferred perspective We use our cognitive map as a guide by aligning images with our environment Scholl (1987) • College students were asked to point in the direction of unseen campus locations • They were better able to do this if the locations were in front of them

28. Organization in Spatial Memory The recall of verbal material is often marked by organization • Items are systematically related or which share pre-experimental associations are recalled together during output • Spatial memory also shows organization as reflected by the presence of organization during output

29. Organization in Spatial Memory Menzel (1973) Procedure • The mental map of young chimpanzees was tested on a 1-acre enclosure • As one researcher walked the chimp around the field, a second researcher (in full view of the chimp) placed pieces of banana and lettuce • The researchers crisscrossed the field distributing the food in random fashion Results • Chimps in searching for hidden food, maximized the rate of food acquisition by using a least-distance strategy • This knowledge of distances was also combined with the ability to measure angles, allowing them to find out the hidden place of food, symmetrically opposed to another one • They also bypassed the lettuce to get the fruit first – showing a priority was put on the rewards

30. The Development of Spatial Memory in Children Cornell & Heth (1979) Procedure • Infants seated in mothers laps with small projection screens on either side • Slides depicting random shapes were projected to one side every 10 seconds and a constant checkerboard pattern appeared simultaneously on the opposite side • Infants orient to novel stimuli and look less at repetitive stimuli so they learned to look at the changing patterns • To test whether the infants had learned turn responses or a cognitive map, the mother turned the chairs around to face in the opposite direction • Therefore, the orientation with respect to the novel versus repeated slides was rotated

31. The Development of Spatial Memory in Children Cornell & Heth (1979) Results • The youngest infants (4 months old) continued to turn in the same direction as before – response learning • The older infants (12 months old) correctly compensated for change of orientation within the room and now turned in the opposite direction – place learning

32. Motor Skills Learning The acquisition of precisely adjusted movements in which the amount, direction, and duration of responding corresponds to variations in regulating stimuli Pursuit Rotor Task • The goal is to keep a stylus on a fixed point on a rotating disk Mirror Tracing Task • The goal is to follow the outline of an object with a pencil or stylus with visual guidance coming from a mirror

33. Motor-skills Learning: Implicit or Explicit? Both implicit (procedural) knowledge and explicit (declarative) knowledge are at times evident in motor-skills learning Sometimes implicit – we can’t really describe our actions • Alzheimer’s patients have no trouble with the mirror-tracing task yet are grossly impaired in declarative memory tasks (recall of word lists, etc.) Sometimes explicit – conscious intention to learn, verbal self-guidance, knowledge of the goal, etc.

34. Amount of Practice Power Law of Practice • The power law of practice is a very general law in human cognition, and in particular in human learning • The higher the level of expertise and the time spent on the task, the more difficult it is to improve (principle of diminishing returns)

35. Schedules of Practice Spaced practice advantage applies to motor-skills learning here too Baddeley & Longman (1987) Procedure • British postal workers practiced typing • Participants were divided into four groups who received either one or two training sessions per day with each session either one or two hours in length See this design next slide

36. Schedules of Practice Baddeley & Longman (1987) How to best use 60 hours of training to maximize performance and learning? 1 Session 2 Sessions 1 hour 12 weeks 6 weeks 6 weeks 3 weeks 2 hour

37. Baddeley & Longman (1978): Results

38. Correct # of Keystrokes as a function of practice distribution Baddeley & Longman (1978)

39. Practice-Independent Learning The relatively permanent change in behavioral repertoire occurring without additional experience Walker (2003) Procedure • Students asked to type a 5-number sequence as fast as they could • After a single session, participants could type this 22 times in 30 seconds • Retesting after 12 wakeful hours or after 12 hours that included their regular amount of sleep

40. Practice-Independent Learning Walker (2003) Results • Retesting after 12 wakeful hours = 24 times in 30s • Retesting after 12 hours (regular sleep) = 27 times in 30s Interpretation • The time spent sleeping was more effective in increasing performance • Certain phases of sleep appear to be important for the long-term consolidation of recently acquired skills and habits

41. Knowledge of Results (KR) Outcome information; feedback • Externally provided information on the success or accuracy of the response that is given to the participant after a practice trial Serves as a basis for corrections on the next trial

42. Knowledge of Results (KR) Thorndike (1931) Procedure • Had students close their eyes while trying to draw 4-inch lines • 3000 lines drawn over 12 sessions • No feedback given Results • No improvement seen Interpretation • Do we need knowledge of results? Yes!

43. Knowledge of Results (KR) Schmidt et al. (1989) Procedure • Participants learned a tracking response; asked to follow a curve projected on a screen that changes in speed and direction • Feedback was either given after every trial, every 5th trial, or every 15th trial Results • Consistent feedback helps during acquisition phase • But after a short 10 minute delay this advantage disappears • Two days later, the group getting the least feedback does the best Interpretation • Is too much feedback detrimental to performance? Yes, it appears that this is true as well

44. Knowledge of Results (KR) Why would less frequent KR lead to better performance? Self-Guidance Hypothesis • Consistent feedback may block the ability to detect one’s own errors • We may become too dependent on external KR for error information • Individuals may be less likely to attend to their own bodily kinesthetic feedback and do not learn to recognize good and poor performance Wulf & Schmidt (1989) • Gradually reducing feedback after acquisition seems to lead to optimal performance

45. Delayed KR Would delaying KR lead to better performance? • Yes - delayed feedback seems to have benefits • It seems that delaying feedback allows performers to develop their own error-detection capabilities without interference from the external KR • However, there is a caveat to this – if individuals are distracted during the delay-until-feedback interval, then performance benefits are lost Swinnen et al. (1990)

46. Implicit Learning Is the process by which knowledge of the structure of a complex environment is acquired largely independent of conscious awareness of specific components of that environment • Learning that appears to occur without awareness or intention to learn and often cannot be described in words what has been learned • “The Cognitive Unconscious” Arthur Reber (born in 1940)

47. Some Implicit Learning Tasks Artificial Grammar Reber (1967) • Devised an artificial grammar learning (AGL) paradigm involving a set of letters instead of words • The grammar determines which letters can follow which other letters Here we are presenting at conference in Chicago in 2013

48. Artificial Grammar Reber (1967) Typical Procedure • Subjects are shown a series of letter strings that follow particular complex rules • Participants are initially not told about the rules • After this “training phase,” participants are told about the existence of rules, and have to then classify the next set of strings into ruleful and unruleful strings (“test phase”) • AGL rules are usually very complex finite state grammar rules

49. Artificial Grammar • Examples of ruleful & unruleful strings: VXVS VXXXS TPTXVS TPTPS • Typical classification performance at test is significantly above chance • Subjects are unaware of their knowledge and cannot verbalize the rules • Reber concluded that participants are implicitly learning the abstract rules of the grammar Ruleful  Unruleful  Reber (1967)

50. Implicit Learning: Serial Reaction Time (SRT) Task Nissen & Bullemer (1987) On each trial a light goes on Just press corresponding button Unbeknownst to subject, sequence of lights is rule governed