1
|
- Michael Lamport Commons
- Harvard Medical School
|
2
|
- It must allow for the comparison of performances of different specie=
s of
animals.
- For example, Parker & McKinney (2000) base their comparisons on
Piaget’s theory of
human performances.
- Such theories may not accurately apply to a wide enough range of
species.
|
3
|
- Because performances are based on multiple dimension of situations a=
nd
organisms, indices must include:
- A way to characterize tasks in terms of:
- Horizontal complexity (how much information is in a task)
- Hierarchical complexity (where in a hierarchy is the task that
underlies such phenomena)
- We propose the General Model of Hierarchical Complexity
- To order tasks
- To explain stages of development
- Measures of something like what has been called g.
|
4
|
- Existing research is used to enumerate domains
- We illustrate how to construct a number of forms of g
- Some across domains
- Some within domains.
|
5
|
- Locurto (2004) found that mice do not show a robust general factor
(i.e., first principal component).
- Geary (2004) has argued that domain-general mechanisms are essential=
for
evolutionary psychologists
|
6
|
- That not much g was found
- He used a battery of tasks that provided evidence of learning withi=
n a
few trials
- He used tasks distinct in terms of
- Motivation
- Sensory modality
- Behavior measured
|
7
|
- The ultimate focus of all of domain-general mechanisms is to support
attempts to gain access to and control of resources
- The social (e.g., mates)
- Biological (e.g., food)
- Physical (e.g., territory)
|
8
|
- The values of g may have increased over evolutionary time as new
organisms developed.
- There is a great need for defining g in a way that captures the
modularity versus generality in a systematic way.
- The modules are similar to the notion of domains.
- The modules are thought to be brain function that is specialized for
tasks in a given domain.
|
9
|
- Posits that tasks can be ordered as to their hierarchical complexity=
- The scale so generated is an equally spaced ordinal scale (Commons &=
amp;
Pekker).
- The Model also may used to measure the stages of animal or human
behavior on this absolute scale.
- Stage of performance has the same number and name as the correspond=
ing
order of hierarchical complexity of the task it correctly completes=
.
|
10
|
- Starting with the standard tasks within the standard domains, we
construct an analogue of g.
- There will be a number=
of
types of measures:
- The highest stage of performance attained in each domain (HS) inclu=
ding
the highest stage in any domain (HHS)
- A form of g that is somewhat akin to human g
- A derived measure of generality of performance, g breadth (gB).
|
11
|
- An animal species may be characterized by the highest stage of
performance observed with any amount of training on its best task se=
ries
(HHS).
- This requires information on what the domains are and what the tasks=
are
within each domain.
- This is the most difficult part because we really do not know how to
classify the domains well.
|
12
|
- But we do know what tasks animals accomplish
- Each task has an hierarchical complexity. The highest stage of perfor=
mance (HS)
is just the highest hierarchical complexity of the task that organis=
ms
in the species correctly address.
- Find the task and domain in which the highest stage of performance
occurs (HHS).
- Note that this is one number on the stage scale that runs from 0 to=
14
|
13
|
- The second index, G, is the average of the highest stage numbers of
performance in each domain (HS).
- This is somewhat akin to human g.
- But this would not separate highest stage from how broad g would be=
.
- The average has advantages of the total g
- Because the average is less sensitive to failing to include a doma=
in.
- Note this average falls on the stage scale that runs from 0 to 14.<=
/li>
|
14
|
- The third index called g breadth (gB), measures how broad their
capability is
- It uses the renormed g that removes the effect of the highest stage.
- One divides the average of the highest stage numbers, obtained in =
the
previous step, g by the top stage of the animal (HHS). This would take away the =
effect
of top stage.
- Then one would have three numbers
- Highest stage (HS)
- Average stage across domains (g)
- Breadth (gB).
|
15
|
- Another form of
“smarts” is within domains.
- This within domain form is like the subtasks within the verbal IQ
tasks.
- The within domain form shows flexibility of stage of performance (f=
S)
within domain.
- Choose the domain and task in which one wants to measure flexibility=
.
- Then find the highest stage of performance, as determined by
hierarchical complexity (HSdomain), on a wide variety of tasks with=
in
that domain.
- Average the stage numbers of the task performances within a domain.<=
/li>
- Divide that by the Hsdomain yielding gdomain Breadth.
- This will be a rational numbers between 0 and 14.
|
16
|
- Highest stage in various domains
- For various animals and humans
- Differences in highest stage in various domains
- This shows qualitativ=
ely
breadth of highest stage
|
17
|
- Animal Prey
defense Predator action
Communication Cohesion
- Spiders 2 2 1 0
- Mice 3 3 2 2
- Pigeons 4 4 4 3
- Crows 5 5 3 3
- Grey Parrots 4 5 5 4
- Spider monkeys 6 6 4 6
- Rhesus monkeys 5 7 5 7
- Chimpanzees 8 8 7 8
- Humans 7-14 7-14 9-14 9-14
|
18
|
- Fish swim to and up the stream where they were born but they do not
perhaps find other pla=
ces
(Order 2).
- Birds migrate using pretty much the same routes year after year. They find their way back to=
their
nests.
- Rat are good at mazes (Order 3).&nb=
sp;
They form concepts but never advance to the nominal stage in =
any
domain.
- Pigeons are particularly good at way finding. They also follow features a=
nd
seem to have a magnetic portion of their brain (Order 4). They reach Stage 4 in conce=
pt naming
|
19
|
- Horses always find their way home. They acquire a much broader amoun=
t of
information, and can be trained to do a wide variety of tasks. They =
are
also quite good at tasks in a number of other domains (Order 3).
- Elephants have large ranges. &=
nbsp;
There ranges are not linear as compared to migrations of birds
and many other animals (Order 4).
- Human way-finding is quite different and more variable.
- Humans make and read maps showing contours and features such as tra=
ils,
roads, cliffs, junctions, etc.&nbs=
p;
(Order 7-12)
|
20
|
- Spiders build webs
- They sense the tension in their legs and squirt the silk (Order 2)<=
/li>
- They wrap their prey in their silk (Order 2)
- But that is one of a relatively small set of actions they engage in=
.
- An 11 year makes a spider web with a syringe filled with a protein l=
ike
that in a spider web
- They do so by imitating the web of the spider. (Order 10)
- They use the part of the same skills to read a map.
- They can do all sorts of other tasks as well.
- Even though they both can build spider webs, and probably the
spider’s is more elegant - the fact is that the human can do so
many things the spider cannot. (Order 10)
|
21
|
- Pigeons can switch which key they use when a few examples from a cla=
ss
are switched.
- They form a large number of arbitrary concepts even including such
abstract examples as inside and outside. (Sensory-motor order 3)
- They name classes such as fish (respond quickly) versus non-fish
(respond slowly). Whe=
n a
few examples are switched so that the class name has been switched =
from
respond quickly to respond slowly, they switch their rate for all
members of the class. (Nominal Order 4)
- Sparrows on the other hand probably cannot do this.
- Crows sequence nominal stage actions by planning to bend a wire to r=
each
around a corner in a plastic tube to get food. (Sentential Order 5).
- African grey parrots sequence words and understand word sequences
distinguishing the difference between the passive voice and active
voice.
- But the African Grey parrots solve all sorts of problems of the
sentential order
- such as saying letters and numbers in order
- counting small sets of objects systematically.
|
22
|
- Note that animals for the most part do not vary much in their stage =
of
performance within species.
- This is not true within
humans.
- Adults range from primary stage 6 to metasystematic stage 12.
- There are also great differences from person to person as to their
relative stage of performance on tasks in different domains.
|
23
|
- We have presented some indexes for assessing how broad the smarts of
animals might be.
- In no way, have we answered the question as to whether their are gen=
eral
mechanism or just separate modules.
- What we have possibly shown is that there is variability in different
species as to
- their breadth of domains at that correctly address tasks of a given
order of hierarchical complexity
- their highest stage.
|