Transcript The cultural origins of colour categories

The cultural origins of
colour categories
Tony Belpaeme
Artificial Intelligence Lab
Vrije Universiteit Brussel
Introduction
•Colour spectrum is continuous…
still we divide it into categories.
•Why do we divide the spectrum into these categories?
Arbitrary division
• Until mid 20th century, colour categories
were by many believed to be arbitrary
– “… an American describing [the spectrum] will
list hues as red, orange, yellow, green, blue,
purple … nothing inherent either in the spectrum
or the human perception … which would compel
its division in this way.”
(Gleason, 1961)
Universalism
• Berlin and Kay (1969) used naming
experiments to extract colour categories
• Subjects marked the focus and extent of
colour terms in a colour chart.
Universalism
• Berlin and Kay noticed a universal pattern in
colour naming of different languages.
• But methodological concerns remained.
Universalism
• The universal character has been hailed by
many and has been reconfirmed by some.
(among others Kay and McDaniel, 1978; Durham, 1991; Shepard, 1992;
Rosch-Heider, 1972; Kay and Regier, 2003)
Mechanisms
• Supposing we accept a certain universality
of colour categorisation, what mechanisms
could underlie this?
– Nativism: genetic makeup.
– Empiricism: interaction with the environment.
– Culturalism: cultural interaction with others.
Nativism
•What mechanisms could
underlie universalism of
colour categories according to nativists?
L
+
–Regularities in human
early visual perception,
especially the opponent
character of colour vision.
(Kay and McDaniel, 1978)
–Regularities in the neural
coding of the brain.
(Durham, 1991)
–Genetic coding of colour
categories. (Shepard, 1992)
_
+
+
L
S
M
+
+
+
R-G
_
+
Y-B
Empiricism
• What do the empiricists have to say?
– Our ecology contains a certain chromatic
structure which is reflected in our colour
categories.
– We learn our colour categories by interacting
with our environment.
(e.g. Elman et al., 1996; Yendrikhovskij, 2001)
– This all happens without the influence of culture
or language.
Culturalism
• And finally, how do culturalists account for
universalism.
– Colour categories are culture-specific.
– They are learned with a strong causal influence
of language and propagate in a cultural process.
(e.g. Whorf, 1954; Davidoff et al., 2001; Belpaeme and Steels)
Three stances
• Three opposing explanations
– Nativism.
– Empiricism.
– Culturalism.
– Of course a blend of two or three positions might
be possible.
Discussion
• The discussion has been held on many
different fronts
–
–
–
–
–
–
Neurology.
Psychology.
Anthropology.
Linguistics.
Ophthalmology.
Philosophy.
• We will tackle the discussion from artificial
intelligence and computer modelling.
Artificial intelligence
• AI allows us to create models of natural
phenomena, of which we then observe their
behaviour.
• Different premises can be implemented in
the models, allowing us to get an insight
into the validity of the premises.
– E.g. traffic modelling.
Reflections on empiricism
• Claim: colour categories are extracted from
the environment, which contains enough
structure to explain universality.
• Procedure
– Collect chromatic data.
– Extract colour categories. For this we use a
clustering algorithm.
– Compare extracted categories with each other and
with human colour categories.
• If the claim is true, we would expect a high
correlation between all extracted categories.
Reflections on empiricism
•The chromatic data
–Three data sets: urban, natural and random.
Reflections on empiricism
• Extracting categories from these data.
Categoriesfrom
fromnatural
urban data:
Categories
data:
Reflections on empiricism
• Quantitative comparison
– 11 categories extracted from natural and urban
data
100
80
60
40
b
20
0
-20
-40
-60
-80
-100
-80
-60
-40
-20
0
20
a
40
60
80
100
Reflections on empiricism
•Comparing with human colour categories
–By computing correlation between extracted
categories and human categories.
–Correlation between extracted categories
Reflections on empiricism
• Intermediate conclusion
– The claim that human colour categories are
specified by the distribution of chromatic stimuli
in the world is not supported by our data.
– However, there does seem to be a twofold
influence by
• The structure of the perceptual colour space.
• The properties of perceptual categories.
Reflections on nativism
• Claim: universal colour categories can arrive
from natural selection.
• Procedure:
– Take a population of simulated individuals, of
which the colour categories are evolved.
– Fitness is defined as how well an individual can
discriminate colour percepts.
• If the claim is true, we would expect an
evolutionary process to produce a repertoire
of colour categories, shared by all
individuals.
Reflections on nativism
• Indeed, all individuals after enough time end up
with identical colour categories
• Evolution keeps track of ecological pressure.
Reflections on nativism
• Some observations
– Colour categories do not seem to be latently
present in the brain. (Davidoff et al., 2001)
– Technologically advanced cultures use more
colour categories than “stone-age” cultures.
• Question
– Is natural evolution fast enough to keep up with
technological evolution?
Reflections on nativism
• In simulation, evolution is not fast enough.
– Adapting to ecological changes takes
approximately 20 generations (≈ 400 years).
• Natural evolution is undirected.
– Achieving a categorical repertoire can only
happen through blind mutations and
recombinations of hereditary material.
– If evolution would be “rebooted” would
humankind arrive at different colour categories?
Reflections on culturalism
• Claim: language and culture have a causal
influence on colour categories, and thus
cause universality.
• If true, we expect linguistic interactions to
cause sharing of colour categories.
• Procedure
– Take a population of simulated individuals that
learn colour categories and communicate about
colour.
Simulating cultural transmission
• The ingredients
– Agent-based simulations
• An agent is a simulated individual, with perception,
categorisation, lexicalisation and communication.
• Perception maps spectral power distribution onto an
internal colour space.
• Categorisation maps percepts onto categories, categories
have prototypical behaviour.
• Lexicalisation connects categories to words.
• Communication takes care of uttering word forms.
• The agents have no way to access the internal state of
other agents: no telepathy!
– Several agents make a population
Simulating cultural transmission
L
a
b
speaker
hearer
=red
red
“Red”? Do I know this
form? If so, is it
uniquely related to a
stimulus?
Reflections on culturalism
• Colour categories of
two agents
5
4
category variance
• Sharing of colour
categories
4.5
3.5
3
2.5
2
1.5
1
0.5
0
0
10000
20000
30000
game
40000
50000
Reflections on culturalism
• Influence of language on categories
25
20
category variance
40
b
20
0
-20
60
40
15
20
b
60
0
10
-20
-40
-40
5
60
60
40
20
0
0
-20
0
2000
60
4000
6000
8000
20
10000
game
L
80
20
12000
0
14000
40
-40
a
40
80
60
16000
-20
-40
a
18000
20
20000
40
L
• But as language is culture-specific, cultural
evolution cannot explain universalism.
Summary
• Empiricism is not a good candidate to
explain universalism
– There is not enough ecological pressure.
• Nativism can explain universalism, but is to
slow to follow ecological pressures.
– Also, recent neurophysiological and molecular
studies point out many differences in colour
perception between individuals.
• Culturalism can explain the sharing of
categories in a culture, but not universalism.
Conclusion
• A blend of all three positions is needed to
explain universalism.
• But language and culture plays a crucial role
as the catalyst which binds the perceptual
categories of individuals.