comparing intelligence Flashcards
what is Anthropomorphism
• Anthropomorphism: we assume that animal cognition is like human cognition (but not as good)
what is Anthropocentrism
• Anthropocentrism: we interpret “advanced” as meaning “more like us”
what is Social/political baggage
• Social/political baggage: we place the kind of human we are at top of the “ladder of nature”
what is scala naturae
natural ladder
example of big brain = higher intelligence
hippocampus and spatial memory
what is “(En)cephalisation coefficient
raito of brain mass to body mass
evidence of qualitative difference in intelligence
Note how the New World Monkeys, Old World Monkeys and Apes seem to follow a different regression line to other mammals, and to reptiles and amphibians.
what did Jerison say about intelligence
• Jerison: Look for deviations from plot of brain mass vs. body mass. Once this is done, we start to get some interesting patterns emerging.
behavioural test of intelligence example
Hebb-Williams maze (sequence of T-mazes)
sophisticated examples of intelligence tests
– Successive reversal
– Learning set
– Probability learning
what is the Serial Reversal Learning: Mackintosh (1974)
Blue and yellow button, press blue to get food. Once the accurqacy reaches 90%, the correct button reverses and the yellow one needs to be pressed
The result? On later reversals the animal makes less errors in acquiring the discrimination.
Madingley sheep example of Serial Reversal Learning:
A spatial version of the task (with maddingly sheep). They can choose to walk either left (food) or right (nothing). They do this task until it gets it 90% right, then switch the right path. After a while, if they get an answer wrong, they instantly go down the other path. This task seems to suit them.
what are learning sets (harlow) 194
Given two stimuli, need to locate right one. Wait til they get 90% accuracy. Then it changes to another stimuli
The result? On later problems the animal makes less errors in acquiring the discrimination. In extreme cases it makes only 1 error! Can we use the rate of acquisition of this problem as an index of intelligence?
learning set across species
rhesus at the top. squirrel and rat near bottom.
what is the problem of contextual vairabels leading to
why we underestimate the learning for animals such as the rat
what is Macphails null hypothesis
- Macphail (e.g. 1982, 2000): there are no cognitive differences between non-human animals. A more refined version: There are no differences amongst non-human vertebrates.
- The only important difference is the emergence of human language. Language training may confer special abilities.
- If we find more sophisticated cognition in e.g. apes than other species, that may be because we can understand better how to test cognition in species that are like us. => “Contextual variables” could be responsible for the observed differences in performance rather than genuine differences in intelligence
the role of contextual vairables with goldfish
In the first condition, the goldfish didn’t learn. The second condition makes the reward appear within a certain tiem frame. By doing this, the goldfish learns the task.
what did Herman and Arbeit (1973) find abotu dolphins
• Herman and Arbeit (1973) found that dolphins had difficulty forming learning sets with visual stimuli but could with auditory stimuli.
what did Herman and Arbeit (1973) conclude
- Thus how well an animal forms a learning set may well depend on the type of stimuli used to test them.
- The only valid test would be to compare animals with similar sensory and effector capabilities (we’ll look at this approach later).
how to end the problem of contextual variabels
• The only valid test would be to compare animals with similar sensory and effector capabilities (we’ll look at this approach later).
evidence in support of Macphail
- Simple forms of learning (e.g. classical and operant conditioning) take place in much the same way and at much the same rate in all vertebrate species and at least some invertebrates
- Surprisingly sophisticated forms of learning turn up in invertebrates, e.g. molluscs, arthropods.
what did Colwill, Absher and Roberts (1988)study
Colwill, Absher and Roberts (1988)
• Conditional discrimination in Aplysia
• Found that Aplysia could learn to provide differential responses to the same stimulus in different contexts.
• Context 1 was a smooth white round bowl with lemony seawater
• Context 2 was a dark grey rectangular container with ridges and turbulence (an aerator).
Conditioning discrimination is shown
who shows conditoning discirminaiton
colwill, abshar and roberts 1988
Learning in honeybees (M. Giurfa & colleagues) stidy
• Classical and instrumental conditioning
• Contextual learning: C1: A+, B-; C2: A-, B+
• Categorization:
– bilaterally symmetrical vs.
asymmetrical,
– different types of abstract patterns,
– same vs. different (matching and non-matching to
sample)
– negative patterning: A+, B+, AB-…w/ olfactory
stimuli
what is bittermans suggestion to the comparabiloty problem
Bitterman’s suggestion: Control by systematic variation of the confounding factors
• Problem: How can we ever explore all variations?
• Another solution is to use comparable species, e.g. pigeons and corvids
what did Wilson, Mackintosh and Boakes (1985) do
- This experiment by Wilson, Mackintosh and Boakes (1985) suggests that jackdaws show better transfer to a novel matching problem after training on matching with different stimuli. Note that in this case the jackdaws actually learn more slowly - but show more transfer.
- There is also some evidence that corvids (crows, rooks, ravens, magpies and jays) perform better at learning sets than pigeons (Mackintosh, 1988).
These graphs show performance of individual birds. On the left is pigeons, right is jackdaws. In jackdaws, the solid lines are pretrained on matching samples are better than those pretained on elsewise.
For the pigeons, there are no evidence on transferring. The worst pigeon was pretrained. The jackdaws show transfer but learn the task way slower than the pigeons.
alternative method of solving transfer probelm
Either, having studied the sample (top middle) which comes on first, 1. pick the bottom left comparison stimulus because it feels more recent (or more familiar), or 2. pick bottom left because that’s the right response to this configuration of stimuli
It could be that Corvids are biased to use solution 1, whereas pigeons tend to use solution 2, but does that make the corvids more intelligent? Note that there is evidence that pigeons can, after a lot of training, use solution 1, so it might simply be that recency or familiarity information is just less salient for them.
recency/familiarity w chimps
Language trained chimps could solve these tasks, but non-language trained chimps could not.
Interestingly, both sets of chimps could solve successive same/different problems, where one object is shown and then either the same or a different object is shown.
e.g. if an object (peg) is taken out, need to notice there is a difference.
When you have these tasks, all of the objects appear recent. The non-language trained chimps are stuck whereas language trained chimps can complete the task.
what did Gillan, Premack and Woodruff 1981 find
Given two bits of a problem and need to identify is they are the same or different. The language monkey can pass this test. This test is used to assess IQ in humans
conclusion on sarah the language learnt monkey
Here we have the conceptual analogical reasoning problem I referred to in an earlier lecture.
Sarah can get these (mostly) right as well. By using two versions with the same alternatives, the experimenters rule out an explanation based on simple preference for one response item.
