Reversal Learning in Rats

1
Reversal Learning in Rats

• Susanne C. Stahl-Bell
• Mark S. Schmidt
• Columbus State University
• Stahl-Bell, S., Schmidt, M.S. (2002, April).
  Reversal learning in rats. Paper presented at the
  1st annual Georgia Undergraduate Research in
  Psychology Conference, Kennesaw State University,
  Kennesaw, GA.

2
Successive Discrimination Reversal (SDR) Learning

• A task in which animals must learn to reverse an
  operant response across successive problems.
• With each reversal, fewer errors are made.

3
Original Problem
Correct response
Reversal
Correct response
4
Past Research

• The SDR task has been used as a measure of animal
  intelligence by some researchers.
• In one study it was shown that there were
  significant differences between species on the
  SDR task (Gossette, Kraus, and Speiss, 1968)

5
Species Differences

• Gossette et al. used a spatial SDR task to
  compare the learning abilities of kinkajous,
  squirrel monkeys, coatis, skunks, raccoons,
  cacomistles, and capuchin monkeys.

6
Raccoon
Squirrel monkey
Coati Mundi
Cacomistle
Skunk
Kinkajou
Capuchin
7
Cumulative Errors Across 19 Reversals (Gossette
et al. 1968)
8
A Previous Study with Rats

• An earlier study by Dufort, Guttman, and Kimble
  (1953) studied SDR in rats.
• All but one rat in their experiment were able to
  learn an SDR task to near perfect performance (1
  error) per reversal.
• However, the rats could have used visual cues in
  addition to spatial cues to solve the SDR task.

9
Our Study

• We also wanted to study rats in a SDR task but
  where the stimuli were identical.
• Our study was more like Gossettes study because
  we used a spatial SDR task with identical visual
  stimuli.

10
Our Study

• We hypothesized that our rats would not do as
  well as Duforts rats because our rats would not
  have distinctive visual cues in the stimuli.
• We wanted to see where our rats would rank in
  Gossettes hierarchy of species.
• How smart is a rat compared with raccoons,
  monkeys, etc.?

11
Subjects

• We used 8 male rats acquired from Harlan Sprague
  Dawley.
• All were from the Dark Agouti (DA) strain, which
  is known to have superior vision.
• The rats were approximately 3 months old and were
  food deprived for 48 hours prior to testing each
  day.

12
Apparatus

• The stimuli were two identical white Styrofoam
  cups, 9.5 x 4.5 x 8 cm.
• The cups were placed in distinct spatial
  locations (left-right) 31 cm apart on a table
  top, 100 cm from the edge of the table.

13
Procedure

• On each trial the rat was released at the edge of
  the table and had to walk to the cups and move
  aside a cup to get a food pellet.

14
Procedure

• Each reversal consisted blocks of 20 trials.
• The rat was considered to have learned the
  reversal if he got 18/20 correct.
• The rat had to complete each set of 20 trials,
  even if they got the first 18 responses correct.
• The rats underwent a total of 15 reversals.

15
Results

• The rats made significantly fewer errors across
  reversals as expected.

16
Mean Errors / Reversal
r -.85 p lt .01
17
Results

• Mean cumulative errors increased at a slower rate
  than that seen in five of the seven species
  tested by Gossette et al. (1968)
• By the 15th reversal our rats had accumulated
  about 100 errors.
• Based on this comparison, our rats ranked between
  cacomistles and raccoons in spatial SDR
  performance.

18
Our Results Compared with Gossettes
19
Cumulative Errors Across 19 Reversals (Gossette
et al. 1968)
Our rats
20
Discussion

• Gossettes study as well as our study uses a
  quantitative approach to study animal behavior.
  This follows the older view that animals can be
  ranked in terms of intelligence for example by
  comparing how well they perform on a specific
  test and how many trials it takes to acquire the
  task. Using this approach, our rats were ranked
  in terms of intelligence between the raccoon and
  cacomistle.

21
Discussion

• Because many problems have been pointed out by
  using the quantitative approach, most researchers
  today say that qualitative approaches to study
  animal behavior should be used.
• It has been pointed out (Thomas, 19??) that
  quantitative approaches dont take sensory
  differences, motor abilities, or motivational
  differences into account.

22
Subjects Not Used