1. Causality
The relationship between cause and
effect. The principle that all events have
sufficient causes.
The differentia (distinguishing properties/characteristics) of causality which all causal relations
have in common:
The relationships held between events, objects or states of affairs.
The first event A (the cause) is a reason that brings about the second event B (the effect)
The first event A chronologically precedes the second event B (in some cases, a simple spatial
or even conceptual separation is accepted: "Tides are caused by the moon", "Day is caused by
the rotation of the Earth", "Lightning causes thunder")
Events like A are consistently followed by events like B
The cue ball causes the eight ball to roll into
pocket."
"Heat causes water to boil."
"The Moon's gravity causes the Earth's tides."
"A hard blow to the arm causes a bruise."
"My pushing the accelerator caused the car to
go faster."
2. CONDITIONING AND
CAUSATION
• Causality detection is vital to
behavioral adaptation in humans.
• Can associative learning principles
in animals teach us something
about how people detect causes?
• Recently, methods and principles
of operant and Pavlovian
conditioning in animals have been
applied to causality detection in
humans.
3. CONDITIONING AND
CAUSATION
• How plausible is this connection?
• In text, we have argued that associative
learning is ―causality detection.‖
• Animals appear to be learning about
causes of important events in world.
• In Pavlovian conditioning, causes are
environmental stimuli.
• In operant conditioning, causes are
organism‘s own actions.
4. CONDITIONING AND
CAUSATION
• How can we evaluate this
account?
• One can directly study human
causal judgment using methods
inspired by animal learning
research.
5. David Hume and Causality
• Familiarity with David Hume‘s
ideas is helpful in order to
comprehend causality detection.
• More than any other thinker, this
18th century Scottish philosopher
has shaped our understanding of
causality detection.
6. David Hume and Causality
General
• Causation is psychological impression.
type of
• Succession of experiences is given;
learning
but, impression of connection goes
across all
beyond sensory evidence.
animals
• For example, lightning regularly
precedes thunder.
• We associate one with the other; but,
we do not sense their interconnection.
Unconsoius, Black box ; A form of learning that
does not relie on logic and reason (language)
Associative modular
8. David Hume and Causality
• Three conditions are crucial to
forming causal impressions:
• 1. Cause and effect must be
contiguous in space and time.
• 2. Cause must occur prior to
effect.
• 3. There must be a constant
connection between cause and
effect.
9. David Hume and Causality
• Three other conditions better define and
sharpen causal attributions:
• 4. Same cause produces same effect,
same effect comes from same cause.
• 5. When several different events
produce same effect, it must be due to
something that events share.
• 6. Any difference between effects of
similar events must arise because
events differ from one another.
10. David Hume and Causality
• Mechanical model of causal
perception:
• Completely non-conscious
processes lead us to automatically
associate consistently contiguous
experiences.
• This learning is too important to
leave to rational thought.
Understanding cause and
effect is essentail for all
organisms to suvive
11. David Hume and Causality
• Hume‘s principles suggest that a single
pairing of cause and effect will not forge
a firm causal association.
• A causal impression rises to its greatest
strength by degrees as a function of
number of cause-effect pairings.
• Consistency of relation affects ultimate
strength of causal association; any
inconsistency weakens connection.
12. David Hume and Causality
• Also relevant to strength of
causality judgment is superiority of
one possible cause above rival
candidates.
• If several events often precede a
given event and if one does so
more reliably than others, then it
will be singled out as the cause.
13. David Hume and Causality
• Three conditions are crucial to forming
causal impressions:
CONSISTENCY
• 1. Cause and effect must be
contiguous in space and time.
• 2. Cause must occur prior to effect.
SUPERIORITY
• 3. There must be a constant
connection between cause and effect. NUMBER
14. David Hume and Causality
• Three other conditions better
define and sharpen causal
attributions:
• 4. Same cause produces
same effect, same effect
comes from same cause.
• 5. When several different
events produce same effect, it
must be due to something that
events share.
• 6. Any difference between
effects of similar events must
arise because events differ
from one another.
CONSISTENCY
NUMBER
SUPERIORITY
15. David Hume and Causality
• Hume‘s ideas are similar to RescorlaWagner model of associative learning:
• Is completely mechanical and phrased
as a simple mathematical equation.
• Posits gradual growth of associations to
an asymptotic level.
• Expects unpaired events to lower
Extincition
strength of associative learning.
• Involves competition among causes. Blocking
16. David Hume and Causality
• Hume hypothesized operation of
same associative principles in
nonhuman animals.
• This testified to the breadth of
these associative principles, plus
their operation in absence of
language or logic.
17. EMPIRICAL INVESTIGATIONS
OF HUMAN CAUSALITY
DETECTION
• Just what is known about causality
detection beyond its suspected
operation in animal conditioning?
• We next explore human research
directed at causality detection in
controlled laboratory settings.
21. Contingency
The 2 x 2 cont ingency table representing th e presence and ab sence of a cause and an effect.
Effect present
Effect absent
Cause present
Cell a
Cell b
Cause absent
Cell c
Cell d
Overestimate A and
B (subjects behave)
22. Contingency
The 2 x 2 cont ingency table representing th e presence and ab sence of a cause and an effect.
Effect present
Effect absent
Cause present
Cell a
Cell b
Cause absent
Cell c
Cell d
Hume and Rescorla argue that
cause and effects are
detemined on a trial by trial
basis
(order of training causes
overestimate)
Experience by experience
Moment by moment
Overestimate A and
B (subjects behave)
Real time
23. Contingency
• General formula: a/(a+b)-c/(c+d)
• Specific experimental
investigation:
• P(Light|Tap) and P(Light|No Tap)
24. Experimental questions
Must ask organinsm the correct
question and provide the
opertunity to answer (behave)
appropriatly
Contingency
• Data suggest that people
can quite keenly detect
prevailing responseoutcome contingency:
positive or negative or zero.
• Under proper experimental
conditions, people can with
great accuracy and little
bias report positive,
negative, and noncontingent
response-outcome
relations.
The 2 x 2 cont ingency table representing th e presence and ab sence of a cause and an effect.
Effect present
Effect absent
Cause present
Cell a
Cell b
Cause absent
Cell c
Cell d
28. Humans not perfect,
Some Bias
Past experience (firing caused tanks
to explode)
Just rated last trial
29. Contingency
What happen last trial +
• Effects of differential conditional
What has happened in the past =
probability and increased training
Future likelyhhod
are exactly what would be
expected if people‘s causality and
Humans more Flexiable
contingency judgments were
Probaliy not just an ‗assocaitive
based on an associative learning
modular
process like that captured in
But cognition,emotion, motivation,
Rescorla-Wagner model.
bias affect learning
• Conditioned inhibition also occurs.
31. Thomas Procedure With
Humans
Subjects could
make outcomes
(points) move
forward in time
and did (cause
and effect) but
doing so
reduced the
total amount of
points
CONTROL
32. Thomas Procedure With
Humans
• Results are like those with rats.
Temporal • Contiguity promotes operant
contiquity can ‗overshadow‘
contingency responding.
in assocaition formation
• Contiguity also leads to more
positive causal ratings, despite no
Hume (superioty)
actual contingency.
• Both results also hold with
negative contingency.
35. Relative validity effect
Differentially (A,B)
Non Differentially
15.7
• Group C: AX+ versus BX• Group U: AX+/- versus BX+/• More control by X in Group U
despite equal association of X
with + and – (at first)
• True of rats, rabbits, and
pigeons in conditioning
experiments
• True of human causal
attributions
36.
37. Blocking
15.8
Predicitive
(learning)
• AX+ alone
Diagnostic (no prior
• A+, then AX+
causality
assumtions)
• Responding to X is greater in first
case for animals in conditioning
experiments
• And, for humans in causal
judgment experiments
38.
39. LEARNING AND COGNITION: A
THEORETICAL PERSPECTIVE
• Studies on human causality
detection are consistent with
Hume‘s theoretical approach plus
literature on Pavlovian and operant
conditioning in animals.
• Concordances are not peculiar to
findings discussed in Chapter 15.
• Also extends to
overshadowing, configural
conditioning, and occasion setting.
40. LEARNING AND COGNITION: A
THEORETICAL PERSPECTIVE
• An additional concordance is that
Rescorla-Wagner model is also a
promising theory of causality
detection in human beings.
• So, basic mechanisms of
association formation may have
just as much to say about causality
detection in human beings as they
do about conditioning in animals.
41. LEARNING AND COGNITION: A
THEORETICAL PERSPECTIVE
• This is as it should be if
fundamental mechanisms of
learning and behavior are truly
general.
• Hume‘s belief in generality of
associative learning and its
centrality to mind and behavior is
thus supported by work we have
reviewed in our course and
textbook.
