17 - Learning about time

Question 1

Time estimation

Answer 1

• We are pretty good at estimating time periods and making judgements about whether intervals are shorter or longer than each other.
• We are also sensitive to the day/nighttime and 24-hour cycle
• E.g. waking up just before your alarm goes off
• And not just because of the sun: jetlag

Question 2

Periodic and interval timing

Answer 2

• Distinguish periodic (learning to respond at a particular time of day) and interval timing (learning to respond after a particular interval of time)
• PERIODIC TIMING: e.g. Circadian rhythms.
• Question: is the cyclical behaviour really controlled by time per se? Or is it controlled by stimuli that are always present at that particular time?

Question 3

Wheel running in the rat (described in Carlson)

Answer 3

• Rats always ran at night, so researchers manipulated light levels. Continued to run throughout the day if they turned the lights off
• What happens in constant dim light when no light cues are available?
• They started the activity at usual times but became later and later overtime
• They maintain behaviour on approx. 25 hr cycle

Question 4

Measuring periodic timing with activity

Answer 4

• Cockroaches (Roberts, 1965). Increased activity at dusk
• When removed visual cues cycle drifted until increased activity started 15 hours before dusk (cycle slightly less than 24 hours)
• Restoring visual cues produced a gradual shift back to correct time
• Entrainment: light acts as a zeitgeber synchronising the internal clock

Question 5

Is the apparent internal 24 hr clock the results of environmental experience or innate?

Answer 5

• Bolles & Stokes (1965):
• Subjects born and reared under either 19-, 24- or 29-hour light/dark cycles
• Then fed at a regular point in their own particular cycle and food delivery signalled a few hours before by a change in lighting
• Results: animals on the 24-hour cycle learned to anticipate food but the others didn’t

Question 6

Physiological basis of periodic timing

Answer 6

• Is there any evidence for a physiological system that could provide this 24-hour clock?
• The suprachiasmatic nucleus (SCN) of the hypothalamus may be a candidate
• The metabolic rate in the SCN appears to vary as a function of the day-night cycle
• Lesions of the SCN abolish the circadian regularity of foraging and sleeping in the rat. Receives direct and indirect inputs from the visual system, which could keep circadian rhythms entrained with the real day-night cycle
• More recent work suggests every cell in the body has a circadian rhythm, which are all under the control of the SCN
• This can dictate e.g. circadian variation in sensitivity of tumours to chemotherapy
• Disruption in circadian rhythms can be responsible for physical illness (e.g. shift workers more susceptible to heart disease, diabetes, infections and even cancer)
• Sleep and circadian rhythm disruption is also associated with several types of mental illness, such as depression, schizophrenia, bipolar illness
• In Alzheimer’s disease the phenomenon of sundowning refers to the worsening of symptoms in afternoon/evening

Question 7

Interval timing

Answer 7

• Normal classical conditioning procedure: tone (20 sec) = food
• Responses per minute increase over sessions
• Measuring interval timing: the peak procedure
• So, what happens if the stimulus keeps on going (and you omit the food)?

Question 8

Church & Gibbon (1982)

Answer 8

• Rats in lit chamber. Occasionally houselights went off, for a 0.8, 4.0 or 7.2 sec (the CS)
• When the lights went on again a lever was presented for five seconds
• If the rat pressed the lever after a 4-sec CS it got food, otherwise it did not
• Then tested with a range of stimulus durations (0.8 - 7.2 secs)
• Results: for food after 2 seconds, response probability peaked at 2 and then massively decreased
• For food after 4 seconds, response probability is a mountain, peaking at 4
• For food after 8 seconds, response probability is a slow increase to 8
• Conclusion: the rats learned not just that food would occur, but also when it would occur (time their behaviour to match when a reward is expected)

Question 9

Weber’s law

Answer 9

• The just noticeable difference when you change a stimulus is proportional to the initial intensity/magnitude of the changed stimulus.
• Hence in absolute terms small amounts judged more accurately than large amounts
• You can tell one from two sugars more easily than eight from nine

Question 10

Weber’s law applies to time

Answer 10

• I / I = k
• I = Just discriminable change (JND = just noticeable difference)
• I = original intensity (of the standard)
• k = constant
• The critical point is that percentage change is more important than absolute change
• (standard – comparison) / standard
• Illustrates Weber’s law: points where responding has dropped by half
• Timed interval short (2s) between ~ 1s, 3s
• Timed interval medium (4s) between ~ 2s, 6s
• Timed interval long (8s) between ~ 4s, 12s
• Reaches half maximum at ~ 50% of duration in each case

Question 11

Scalar timing theory

Answer 11

• Pacemaker – Working memory – Reference memory = comparator = response?
• Pacemaker emits pulses at a roughly constant rate t (there is random variation)
• When a stimulus is presented, a switch is operated, and the pulses are allowed to accumulate in working memory. This will equal t multiplied by the number of seconds that have passed (N)

Question 14

How scalar timing theory works: process 1

Answer 14

• Storing duration of a stimulus in STM)
• Pacemaker: T = 1 per sec
• Working memory: N x T
• 5-second stimulus: successive pulses stored in working memory

Question 15

How scalar timing theory works: process 2

Answer 15

• Short duration of a stimulus in reference memory
• When the reinforcement occurs, pulses stop accumulating; another switch allows the number of pulses in working memory (N * t) to be stored in reference memory
• This storage is not completely accurate – there is some memory distortion
• This is represented by K, a number that is close to 1:
• If K = 1 the memory is accurate
• If K < 1 a smaller number of pulses is stored
• If K > 1 a greater number is stored
• Error is proportional to duration scalar again
• After several trials there will be several numbers stored in reference memory Nm1, Nm2, Nm3, etc – each equal to the K * N * t for that particular trial
• Remember the error on each trial will not be the same

Question 16

How scalar timing theory works: process 3

Answer 16

• Using stored value in reference memory to decide whether to not to respond on the next trial
• On each trial the animal compares the number of pulses in working memory (N * t) with a random value drawn from those stored in reference memory Nmx
• This is done by the comparator. If the values are close, then the animal responds
• Another stimulus occurs, and the successive number of pulses is stored in working memory
• The animal uses ONE of the values in reference memory to decide when to respond
• The comparator works out how close the values are using a ratio rule – NOT
• a difference rule
• i.e. NOT Nt -NMx but Nt -NMx / NMx
• This is one of the reasons that accuracy is better with short intervals
• Large therefore don’t respond, small therefore respond

Question 19

Problems with scalar timing theory: Conditioning and timing supposedly occur at the same time, and yet are controlled by completely different learning mechanisms

Answer 19

• Some theories of timing try and explain conditioning
• Calculate rate of reinforcement during stimulus, and rate of reinforcement during background
• If rate of reinforcement in CS is higher than rate of reinforcement in background = conditioning
  a) 6 reinforcers in 60 minutes of background = 1/10 = 0.1 foods per minute
  b) 4 reinforcers in 15 minutes of stimulus = 4/15 = 0.27 foods per minute
• This theory cannot explain basic phenomena, like blocking.
• Some conditioning models try to explain timing – e.g. Real time models
• They work just like regular conditioning theories. However, the stimulus is assumed to change over the course of its presentation, and this allows the animal to learn about when a reinforcer occurs

Question 19

Problems with scalar timing theory: There is as yet no physiological evidence for a pacemaker

Answer 19

• Alternatives have been proposed:
  a) Instead of a pacemaker, it has been proposed that timing could be achieved by a series of oscillators, each of which has two states, on or off
• If each oscillator switches after a different period of time, then the entire pattern of activation could be used to determine the exact time
  b) Another solution that has been proposed is the Behavioural theory of timing
• When the animal gets a reward, this stimulates behaviour
• The animal moves across an invariant series of behavioural classes in between reinforcements
• A pulse from an internal pacemaker will change the behaviour from one class to another
• The behaviour that is occurring when the next reinforcer occurs becomes a signal for that reinforcer