Innovation, Fatal Accidents, and the Evolution of General Intelligence

1
Innovation, Fatal Accidents, and the Evolution
of General Intelligence

• Linda S. Gottfredson
• University of Delaware
• Lunch seminar at UC Davis
• Department of Psychology, Psychobiology Group
• December 6, 2005

2
(No Transcript)
3
(No Transcript)
4
Evolution of Division of Labor
Applicants for Attorney, Engineer Teacher, Programmer Secretary, Lab tech Meter reader, Teller Welder, Security guard Packer, Custodian 80 100 120 IQs Middle 50 108-128 100-120 96-116 91-110 85-105 80-100
5
Humans Remarkable Intellect
Encephalization quotient (EQ) brain-to-body
size compared to the average mammal
EQ
6
5
4
3
2
1

MYA 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 .5 .1
Homo sap. sap.
Homo sapiens
Homo erectus
Homo habilis
FIRE
Australopithecines
Chimp
6
The Explanadum

• Human Intelligence
• Psychometric viewg
• General ability to learn reason
• General (cross-domain) utility
• Instrumental (not socioemotional)
• Evo Psych viewsvaried, but mostly not g
• Modular Narrow, domain-specific, automated
• (many fast and frugal heuristics)
• Social intelligence (not ecological
  competence)

E.g., Fitness signaling survival theories
consistent with g
So, most Evo Psych theories leave g unexplained.
7
What is g?

• All mental tests measure mostly the same ability
  g
• g is the spine or core of all mental abilities

g
IQ
V
Q
S
M
others
8
g Mental Manipulation

• Concrete Example
• Digits Subtests
• Forward vs. Backward

Illustrates differences in task complexity More
complex more g loaded
9
gLearning Ability (Typical Learning Needs at
Different IQs)
Written materials experience
Mastery learning, hands-on
Learns well in college format
Very explicit, structured, hands-on
No. of people
Can gather, infer information on own
Slow, simple, concrete, one-on- one instruction
70 80 90
100 110 120 130

IQ
MR
MG
10
g Problem Solving
11
g Plan, Anticipate Problems
12
Performance More Dependent on g in More Complex
Jobs
Applicants for Attorney, Engineer Teacher, Programmer Secretary, Lab tech Meter reader, Teller Welder, Security guard Packer, Custodian 80 100 120 IQs Middle 50 108-128 100-120 96-116 91-110 85-105 80-100
.8 .5 .2
Predictive validity of g
13
Even simple jobs too complexfor some people
Urban hospital outpatients diabetics not knowing that Health literacy level Health literacy level Health literacy level
Urban hospital outpatients diabetics not knowing that V-low Low OK
Signal Thirsty/tired/weak usually means blood sugar too high 40 31 25
Action Exercise lowers blood sugar 60 54 35
Signal Suddenly sweaty/shaky/hungry usually means blood sugar too low 50 15 6
Action Eat some form of sugar 62 46 27
14
What Must an Explanation of g Specify?

1. Cross-domain value (common cognitive demands
   across different task domains in Homo ecology)
2. Differential impact on survival (g-related
   differences in task performance must create
   g-related differences in survival/reproduction)
3. Ecological demands that are unique to genus Homo
4. Conditions that accelerated selection for g in
   Homo sapiens

Need to lay out a nitty-gritty selection walk
15
Natural Selection, or Sexual Selection?

• My focus here on natural selection
• I.e., external, physical environment matters
• Sexual selection for g may also operate, but is
  not plausibly the whole answer
• Why would it select so strongly for g only among
  humans?
• What would trigger runaway selection for g?
• What about all those individuals who die before
  reproductive age?

16
1. Ecological DemandsHow General?

• Clues from analyzing modern jobs
• Cognitive complexity is major distinction
• Example Judgment Reasoning Factor
• Deal with unexpected situations
• Learn recall job-related information
• Reason make judgments
• Identify problem situations quickly
• React swiftly when unexpected problems occur
• Apply common sense to solve problems

None of these is domain-specific.
17
But wasnt life simpler in the early human EEA?

• Yes, but it was never g-proof
• Opportunity to learn reason within-group
  variation in g opportunity for selection
• Tiny effect size many generations big shift
  in distribution

18
2. g-Related Mortality During Reproductive Years
(15-44)?

• Major cause of death today Fatal injuries
• Mostly unintentional (not homicide or suicide)
• Burns, drowning, vehicle collisions, cuts,
  crushing, falls, poisons, animal bites, exposure,
  etc.
• Many males killed in work-related activities
• True worldwide
• Accident prevention is highly cognitive process.
• Spotting and managing hazards makes same demands
  as do complex jobs (e.g., dealing with the
  unexpected)
• Absolute risk of accidental death is low but
  relative risk is high for lower-g populations
• Imagine death rate is .001 overall, but .003 for
  low g

19
Accident Prevention Also Resembles Complex Jobs
Complex jobs require you to r with complexity Complex jobs require you to r with complexity
Learn and recall relevant information Reason and make judgments Deal with unexpected situations Identify problem situations quickly React swiftly when unexpected problems occur Apply common sense to solve problems Learn new procedures quickly Be alert quick to understand things .75 .71 .69 .69 .67 .66 .66 .55
20
Example Motor Vehicle Deaths
Australian veterans followed to age 40 Death rate per 10,000
IQ above 115 51.3
100-115 51.5
85-100 92.2
80- 85 146.7

• People with lower IQ may have a poorer ability
  to assess risks and, consequently, may take more
  risks in their driving.

2x
3x
But in the EEA too?
21
Non-Warfare Deaths USA vs. Pre-Contact
Hunter-Gatherers
USA (1986) USA (1986) USA (1986) USA (1986) Ache (lt1971)
Age 15-24 25-34 35-44 45-64 15-59
Illness 22 44 72 93 49
Accident 51 31 15 4 37
Suicide 13 12 7 2 0
Homicide 14 13 6 1 14
22
Cause of Ache Deaths (N, lt1971)
Age 0-3 0-3 4-14 4-14 15-59 15-59 60 60
Sex F M F M F M F M
Illness Congenital/degenerative Childbirth 9 1 3 26 2 2 3 4
Accident jaguar/snake lightning lost drowned/falls/other 6 4 1 1 23 19 2 1 1 4 1 3 3 3
Homicide sacrificed with adult homicide/neglect buried alive/left behind ritual club fights non-sanctioned murder 4 2 2 7 6 1 1 1 4 2 2
23
Cause of Ache Deaths (N, lt1971)
Age 0-3 0-3 4-14 4-14 15-59 15-59 60 60
Sex F M F M F M F M
Illness Congenital/degenerative Childbirth 8 7 9 1 3 26 2 2 3 4
Accident jaguar/snake lightning lost drowned/falls/other 1 1 10 3 3 3 1 6 4 1 1 23 19 2 1 1 4 1 3 3 3
Homicide sacrificed with adult homicide/neglect buried alive/left behind ritual club fights non-sanctioned murder 14 10 3 1 3 1 2 4 2 2 7 6 1 1 1 4 2 2
Most are mistakes (faulty minds eye) during
provisioning
Mistakes reverberate
24
Cause of Ache Deaths (N, lt1971)
Age 0-3 0-3 4-14 4-14 15-59 15-59 60 60
Sex F M F M F M F M
Illness Congenital/degenerative Childbirth 19 8 17 11 8 7 9 1 3 26 2 2 3 4
Accident jaguar/snake lightning lost drowned/falls/other 1 1 2 1 1 1 1 10 3 3 3 1 6 4 1 1 23 19 2 1 1 4 1 3 3 3
Homicide sacrificed with adult homicide/neglect buried alive/left behind ritual club fights non-sanctioned murder 26 7 17 2 26 4 18 4 14 10 3 1 3 1 2 4 2 2 7 6 1 1 1 4 2 2
25
Cause of Ache Deaths (N, lt1971)
Age 0-3 0-3 4-14 4-14 15-59 15-59 60 60
Sex F M F M F M F M
Illness Congenital/degenerative Childbirth 19 8 17 11 8 7 9 1 3 26 2 2 3 4
Accident jaguar/snake lightning lost drowned/falls/other 1 1 2 1 1 1 1 10 3 3 3 1 6 4 1 1 23 19 2 1 1 4 1 3 3 3
Homicide sacrificed with adult homicide/neglect buried alive/left behind ritual club fights non-sanctioned murder 26 7 17 2 26 4 18 4 14 10 3 1 3 1 2 4 2 2 7 6 1 1 1 4 2 2
NOTE Many Ache died before mating age Many
evolutionary two-fers child killed after
parent dies
26
What Killed Differentially by g Level?

• Not the obvious
• Not high-interest, high-probability threats to
  bands survival (e.g., starvation, harsh climate)
• Because the fruits of competence are shared
  (e.g., meat from hunting)
• But the minor side-effects of core tasks
• Myriad low-probability, chance-laden, oft-ignored
  risks in daily chores (e.g., accidental injury)
• Costs of injury not shared widely

Recall Spearman-Brown Formula for test
reliability Low-g items can yield high-g test
when many items cumulated (here across tasks,
individuals, generations)
27
3. What Unique to Human EEA?

• Not
• Tool use
• Hunting
• Being hunted
• Climate
• Social living

28
3. What Unique to Human EEA?

• Human Innovation
• Changed physical environment or how humans
  interacted with it (e.g., fire, weapons)
• Improved average well-being but created novel
  risks (e.g., burns/scalds, inattention to snakes)
  
• Put a premium on independent learning and
  foresight,
• especially for recognizing hazards and preventing
  accidental injury and death during core
  activities

Innovation hazards require a minds
eyeimagination, foresight
29
4. How Did Innovation Accelerate Selection for g?

• Five possible accelerators
• Double jeopardy
• Spearman-Brown pump
• Spiraling complexity
• Contagion of error
• Migration ratchet

30
Double Jeopardy
g level
Risk of benefit

Sharing
Risk of injury
31
Social Intelligence View?
g level
Risk of benefit

Sharing

Machiavellian

exploitation
Risk of injury
32
High-g innovators make like difficult for
everyone else
33
Migration Ratchet
Imaginators
Mean IQ rises

• Innovate to adapt to harsher
• climates
• clothing, shelter
• storage, preservation

Relative risk steepens
Bigger consequences More hazards More
complexity More innovations
34
Migration Ratchet
Imaginators
Consistent with mean differences in IQ, brain
size, and skeletal robustness by race/latitude

• Innovations to cope with harsher climates
• clothing
• shelter
• storage/preservation

Relative risk steepens
Bigger consequences More hazards More
complexity More innovations
35
Gene-Culture Co-Evolution of g
g rises

• Not this
• But this

g rises
Biophysical environment
Humans modified their EEA, which modified them.
36
Thank you.

• In press
• Available at
• www.udel.edu/educ/gottfredson