Chapter 11 – Comparative Cognition I: Memory Mechanism
What is comparative cognition?
• Comparative cognition can lead to actions that cannot be explained on the basis of external stimuli and individual
experiences at a time.
Example: thinking you left your apartment unlocked and returning to check it
• Some controversy over what the domain of comparative cognition should be.
• Advocates for cognitive ethology claim that animals are capable of conscious thought and intentionality. The claim is based
upon the complexity, flexibility and cleverness of various examples of animal behaviour. This argument has been debated
and rejected by philosophers for centuries.
• Recent studies show the limitations of conscious intent as adequate explanation of human behaviour
• In contrast to cognitive ethologists, experimental psychologists use the term comparative cognition in a more restricted
- argued that “objective evidence can’t give a demonstration either of the existence or of the non-existence of consciousness,
for consciousness is precisely that which can’t be perceived objectively” and “no statement concerning consciousness of
animals is open to refutation by observation and experiment”
• Contemporary experimental psychologists tie cognitive mechanisms closely to behavioural predictions. Thus, cognitive
inferences can be refuted by experimental evidence. They are careful to accept only those hypothesized cognitive processes
that lead to unambiguous behaviour predictions.
• For experimental psychologists, comparative cognition does not imply anything about awareness, consciousness or verbal
• Comparative cognition refers to theoretical constructs and model used to explain aspects of behaviour that can’t be readily
characterized in terms of simple S-R or reflex mechanisms. Critical feature: employs the simplest possible explanations that
are consistent with data.
• C. Llyod Morgan: pointed out that casual inferences about the rich mental life of animals (which are projections of our own
thoughts, emotions and intentions) hamper knowledge of comparative cognition because they are impossible to prove and
they prejudge the conclusions that we may arrive at through more systemic observations.
Animal memory paradigms
• One of the largest areas of comparative cognition is the study of memory.
• Memory: instance in which an organism’s current behaviour is determined by some aspect of its previous experience.
• Anytime an animal’s behaviour is determined by past events, we can conclude that some type of memory is involved.
• Learning is not possible without memory.
• Comparison of Learning (L) & Memory (M) Experiments:
- Acquisition: initial stage of learning.
(L) Deal with kind of information we acquire and the ways in which we acquire it therefore involve manipulations of the
conditions of acquisition.
(M) Only of interest to the extent that it is relevant to the other two phases
- Retention interval: the time between acquisition of information and a test of memory for that information
(L) Always fairly long because short-term changes in behaviour are not considered to be instances of learning
(M) Varied to determine how the availability of the acquired information changes with time. Have retention intervals of
- Retrieval: the recovery of information from a memory store
(L) Due to emphasis on the other two phases retrieval is kept constant
(M) Dependent on different conditions & situations of retrieval necessary
Phase Studies of Learning Studies of Memory
Acquisition Varied Constant
Retention Constant (long) Varied (short & long)
Retrieval Constant Varied
- Studies of learning and studies of memory all involve the three phases.
• Schachter & Tulving: identified 5 types of human learning and memory
- Procedural memory, perceptual memory, semantic memory, primary or working memory, and episodic or declarative
memory. Not all have animal forms.
- Procedural memory: Memory for learned behaviour and cognitive skills that are performed automatically, without the
requirement of conscious control, often reflecting knowledge about invariant relationships in the environment, such as
CS-US contiguity (classical conditioning) or response-reinforcer contiguity (instrumental conditioning)
1 - Episodic memory: memory for a specific event or episode that includes information about what occurred and when and
where it took place, as contrasted with memory for general facts or ways of doing things.
Working and Reference Memory
• Walter S.Hunter: earliest experimental investigations of animal memory
- tested rats, dogs and raccoons in simple memory tasks
- Apparatus included three boxes with only one being the goal box (marked by light) that had food which was varied from
trial to trial
- After animals learned to always choose box with light difficulty was increased by the light marking of the goal box
remained on for only a short time. After the light was turned off the animals were detained for various lengths of time
and they had to somehow remember which light was on to find the food.
- The longer the animal was delayed the more likely it was to make a mistake as to where the food was. The maximum
time that the rat could handle was about 10secs, the raccoon was about 25secs & dogs didn’t deteriorate past 5 minutes.
- Animal behaviour varied while they were detained.
- Working memory: Temporary retention of information that is needed for successful responding on the task at hand but
not on subsequent (or previous) similar tasks.
- Reference memory: Long-term retention of background information necessary for successful use of incoming and
recently acquired information.
Delayed Matching to Sample
• Delayed-matching-to-sample procedure: participants are reinforced for responding to a test stimulus that is the same as a
sample stimulus that was presented some time earlier. Most versatile technique available for the study of working memory.
• Izumi &Kojima: asked whether chimps can identify a vocalization based on just visual features of a chimp making that
Result was that chimps can also read lips like humans.
Procedural Determinants of Delayed Matching to Sample
• Several aspects of the matching-to-sample procedure are critical in determining the accuracy of performance.
• Important factor is the duration of exposure to the sample stimulus at the start of the trial and the delay interval after the
• Grant’s: tested pigeons in a Skinner box that had three pecking keys in a row on one wall above food hopper. The stimuli
were colours (red, green, blue & yellow) which could be projected on the pecking keys.
- Results indicated that accuracy in the delayed-matching-to-sample procedure decreased as a function of the delay
interval and increased as a function of the duration of exposure to the sample stimulus.
• Trace decay hypothesis: exposure to a stimulus produces changes in the nervous system that gradually and automatically
decreases after the stimulus has been terminated. The stronger the trace, the stronger the effect of the past stimulus on the
• Sargisson & White: tested to see if better memory could be trained by using longer delay intervals from the beginning of
- One group of pigeons the delay between the sample and choice stimuli was always 2 seconds during training. For other
groups, this delay was 4-6 seconds. For comparison, a control group was trained with the usual procedure of presenting
the choice alternatives immediately after the sample
- Result showed that forgetting functions don’t directly reflect the decay or fading of memory for the sample stimulus as a
function of time. Rather, test performance depends on the similarity between the conditions of testing and the conditions
- The common finding that memory gets worse with the passage of time may simply reflect the fact that participants don’t
have practice with longer delay intervals.
• The delay interval used in training is just one training variable that influences delayed matching to sample performance.
• Matching is basically instrumental-choice behaviour motivated by the reinforce provided at the end of the trial. Therefore, it
shouldn’t be surprising that various factors that influence choice behaviour are also relevant in the analysis of the matching
to sample performance.
Response Strategy in Matching to Sample
• The matching-to-sample procedure is analogous to a discrimination problem in that the participant has to respond to the
correct stimulus and refrain from responding to the incorrect one to get reinforced.
• In discrimination training, participants can make the correct choice by focusing on the correct stimulus by inhibiting
behaviour to the correct stimulus, by inhibiting behaviour to incorrect stimulus or by using both these response strategies. In
contrast, participants in matching to sample appear to focus primarily on the correct choice.
• Pigeons focus on the correct alternative, if they saw the matching stimulus they would peck at it without checking was
presented by the other key.
General VS Specific Rule Learning
2 • Most matching-to-sample procedures can be solved either by learning a general same-as rule, or by learning a series of
specific stimulus-response relations.
• Specific stimulus-response learning should not facilitate performance with new stimuli, because the required stimulus-
response association hasn’t been learned yet the first time subjects are presented with novel sample stimulus.
• General rule learning predicts considerable positive carryover, because the general same-as rule can be used to solve any
matching-to-sample problem. Therefore, in tests of transfer from one matching-to-sample problem to another, general-rule
learning should produce better performance than specific-rule learning.
• Oden, Thompson & Premack: first provided chimps with training on a matching-to-sample task with just one pair of
stimulus objects, a stainless steel measuring cup and brass bolt lock.
- Results showed that chimps seemed to have a general same-as rule with just two training stimuli
• Grant’s: only four different colours served as samples, is likely to favour the learning of specific stimulus-response relations.
• In contrast to Grant’s procedures that employ a couple of hundred possible samples are likely to favour the learning of
• Trial-unique procedure: A matching-to-sample procedure in which different sample and comparison stimuli are used on
Spatial Memory in Mazes
• The matching-to-sample procedure can be adapted to investigate how animals and people remember a variety of stimuli
• This section focuses on applications of memories of places
The Morris Water Maze
• Common procedure on spatial memory
• Typically used with laboratory rats or mice
• Circular tank filled with water high enough to force the subjects to swim. Rats and mice don’t like to swim but they are
motivated to because they want to find the platform which isn’t visible as they swim around.
• Results showed that learning progressed fairly rapidly. As they learned the task, subjects took less time to find the platform
and took more direct routes to the platform. Greatest improvement was from the first to second day of training.
The Radial Arm Maze
• More ecologically valid laboratory technique for the study of spatial memory rats and mice.
• Developed to test memory for places where animal recently obtained food and depleted that food source.
• Takes advantage of evolved strategies for finding food in an animal’s environment.
• In many environments, once food had been eaten at one location, it is not available there again for some time until it is
replenished. Therefore, animals have to remember where they last obtained food and avoid that location until the food is
• First developed for rats but analogous procedures have been developed for other species.
• Olton: results showed that during the first five test runs after familiarization with the maze, the rats made a mean of nearly
seven correct choices during each test.
- With continued practice, the mean number of correct choices was consistently about seven, indicating that the animals
rarely entered an arm they had previously chosen on that trial.
• Tendency to avoid recently visited arms in a maze whether or not the maze arms are baited with food at the start of the trial.
This suggests that radial maze performance has deep evolutionary roots.
• Rats are able to remember 16 to 24 spatial locations in a food-depletion working memory task. It is claimed that humans
working memory has the capacity for 7 items give or take 2.
• Spatial memory is not permanent, as time increases error also increases.
BOX: LEARNING AS SYNPATIC PLASTICITY: LONG-TERM POTENTIATION (LTP) AND LONG-TERM DEPRESSION
• Estimated that your brain contains more than 100 billion neurons and each neuron may be coupled through chemical synapse
to thousands of other neurons, forming a complex neural network.
• Donald Hebb: argued that neural connections are strengthened when two cells are active at the same time, an idea that is
sometimes summarized with the mnemonic “cells that fire together wire together”
• Bliss & Lomo: provided neural evidence for Hebb’s experiment.
- Examined how neural activity within hippocampus alters the strength of synaptic connections.
- Long-term potentiation (LTP): if strong input provided, one that caused a very strong response in the mossy fibers,
subsequent inputs produced a much larger response
- Long-term depression (LTD): weakens synaptic connection
- LTP & LDP can both lasts for days to weeks
3 • Researchers have also shown that LTP occurs in the spinal cord. Heightened pain appears to reflect a form of LTP, a cellular
memory within the spinal cord that heightens the pain signal sent to the brain.
Acquisition and the Problem of Stimulus Coding
• Memory depends on our having experienced an event and having made some kind of record of that experience. However,
even when our memory is excellent, it is not because we retain a perfect or literal record of the earlier experience.
• We do not have a literal record of our past experiences in memory. Rather, our experiences are coded in the nervous system
in some way for the purposes of retention.
• Stimulus coding is a critical feature of the acquisition phase of memory which is how a stimulus is represented in memory.
Cognitive Maps and Other Navigational Codes
• The idea that animals form a cognitive map which then guides their spatial navigation has been a prominent hypothesis for
many years, but it calls for further specification. Potentially, maps can represent a variety of different types of information.
To claim that animals form a cognitive map doesn’t tell us precisely what information is contained in such a map and how
animals use that information.
• Beacon following is a fairly simple navigational tactic, requiring little more than the formation of an association between the
beacon and the goal object. Example: sign-tracking from chapter 3.
• Landmarks are a distinctive stimulus that is not at the goal location but has a fixed relation to the goal.
• Geometric features such as those that constitute a triangle provide additional cues for spatial localization.
• Spatial gradient is another source of