Stimulus Control of Behaviour Stimulus Control Differential responding
Stimulus Control of Behaviour
Stimulus Control • Differential responding and stimulus discrimination • Complex environment • “Signal” from the “noise” • What is important?
Reynolds (1961) Train Test Responses per minute • Pigeons, operant chamber, compound key Bird 107 triangle Bird 105 circle
Discrimination and Generalization • Differential responding – Stimulus discrimination • Similar responding – Stimulus generalization
Generalization Gradients • Train on one (or more) stimulus • Test on continuum (i. e. , a “gradient”) of other, related stimuli • Plot on graph… Generalization Gradient • Tricky – Reinforcement problem – Extinction problem – Probe trials and extinction blocks
Responses Gen. Grad. : Stimulus Control Training stimulus Stimulus Continuum (e. g. , wavelength of light)
Responses Gen. Grad: No Stimulus Control Training stimulus Stimulus Continuum (e. g. , wavelength of light)
Spetch & Mondloch (1993) • • • Pigeons in localization task Touch-screen Peck at target area of 2 cm 2 Target defined by landmarks Tested to determine the landmarks that controlled the discrimination
Training Stimuli
Results: Pigeon C 207
Results: Pigeon 271
Factors in Stimulus Control
Sensory Capacity • Organism’s sensory capacity determines what stimuli can control behaviour • E. g. , ultraviolet stimuli for honeybee, but not for human • Physical orientation
Overshadowing • Competition among stimuli for access to learning processes • Higher intensity stimulus more easily conditioned • Speed of conditioning
However… • Stimulus element approach – Elements in a compound stimulus treated as distinct and separate components – Overshadowing: most salient stimulus gains control • Configural-cue approach – Compound stimulus treated as an integral whole – Overshadowing: generaliztion decrement
Group Training stimuli Test stimulus Generalization from training to test Overshadowing weak-STRONG weak Decrement Control weak No decrement • Configural-cue interpretation – Overshadowing due to degrees of generalization decrement from training to testing – No generalization decrement for control group – Overshadowing group learned weak-STRONG, but only tested on weak, so considerable generalization decrement • Evidence for both stimulus element and configural-cue approaches
Type of Reinforcement • Biological predispositions; belongingness • Species • Sometimes seasonal – E. g. , stimuli associated with courtship disregarded outside of mating season
Force & Lo. Lordo (1973) Tone & light Light Tone – L/T … press pedal … food • Group 2 – L/T … press pedal … avoid shock • Test compound and individual components Pedal Presses • Training: • Group 1 (food reinforcer) Group 2 (shock avoidance)
Instrumental Response Factors • Nature of the response required for reinforcement can affect stimulus control
Dobrzecka, Szwejkowska & Konorski (1966)
Learning Factors in Stimulus Control • Pavlov – Inherent properties of nervous sytem – Innate – Due to physical similarity of stimuli • Lashley & Wade (1946) – Explicit discrimination training required – Stimulus control learned – Stimulus generalization because animals have NOT learned the difference between stimuli
Stimulus Discrimination Training • In classical or instrumental conditioning – CS+ or CS-, S+ or S- Cummulative responses • Stimuli explicitly associated with other stimuli or outcomes S+ S- Time
Training and Stimulus Control • Use generalization gradient to determine degree of stimulus control • Need to determine the feature(s) of the discrimination procedure that controls the gradient
Jenkins & Harrison (1962)
Range of Discriminative Stimuli • Wide range of stimuli have been used in discrimination studies – Music, auditory frequencies, painting styles, geometric shapes, etc. • Can use these discrimination studies to assess the sensory capability of species
Pigeon Art Appreciation • Watanabe et al. (1995) • Trained pigeons to discriminate between Monet and Picasso • Achieved high degree of accuracy • Pigeons generalized to other artists of the same style (Impressionist or Cubist)
What’s Learned • Spence’s (1936) theory • Learn about both S+ and S– Stimuli treated separately • S+ represents excitation • S- represents inhibition • Test – Excitatory stimulus generalization gradient • Most response near S+ – Inhibitory stimulus generalization gradient • Least response near S-
Honig et al. (1963)
S+/S- Interactions • Interactions likely to occur between S+ and S • Especially likely if using intradimensional discrimination – Stimuli from same stimulus continuum • Example seen in peak shift
Peak Shift • Maximum responding on generalization gradient not to trained stimulus • With S+/S- training • Already seen in Jenkins & Harrison (1962) Peak shift
Hanson (1959
Spence’s Interpretation • Opponent process system • S+ generates internal excitatory gradient • S- generates internal inhibitory gradient • Actual generalization gradient (i. e. , measured behavioural response) due to net sum of excitatory and inhibitory gradients
Actual generalization gradient (i. e. , net sum) excitatory inhibitory e. g. , responses Hypothetical internal gradients S- S- S+ S- S- Stimulus continuum S-
Absolute/Relative Control • Absolute interpretations – Learn specifics of individual stimuli – Spence’s theory is an absolute theory • Relational interpretatsions – Learn relationship between stimuli
Transposition Task • Kohler (1939) • Can a relational rule be transferred to a new stimuli set? Training S+ S- Testing transfer (relational) absolute
Intermediate Size Problem • Gonzales, Gentry & Bitterman (1954) • Chimpanzees, visual display of squares • Chimps choose intermediate size on transfer test Training 1 4 S+ Transfer Test 9 4 7 9
Training Technique • Simultaneous discrimination training – S+ and S- presented together • Successive discrimination training – S+ and S- presented alone on different trials • Relative stimulus control with simultaneous • Absolute stimulus control with successive
Stimulus Equivalence Training • Train subjects to treat dissimilar stimuli as similar • Training to generalize, not discriminate • Categorization learning
Herrnstein, Loveland & Cable (1976) • • • Pigeons Presence/absence pictures S+: item present S-: item absent Various S+ stimuli: water, trees, people
Stimuli for Water as S+ Condition S- S+
Results Trees Water People
Contextual Cues • Context cues can exert stimulus control • Perform behaviours appropriate to a given context
Siegel (1975) • Morphine tolerance • Home room and injection room • Same amount of morphine across 4 days • Conditional compensatory response to context cues of injection room
Perkins & Weyant (1958) • Two groups of rats run through two mazes, one white, one black; same maze layout • Half of each group tested in same colour maze, half in opposite colour maze • Poor performance for rats tested in opposite compared to same
Kamin (1957) State-dependent learning Rats; avoidance learning Test at various retention intervals Rats’ own internal physiology serves as context cue Avoidance (%) • • 100 50 0 12 24 36 48 60 72 Retention Interval (hr) 84
Akins (1985) • Male quail sexual conditioning • Arena with two compartments – Sand floor, orange walls – Wire-mesh floor and walls, green ceiling • Individual subjects allowed to move back and forth in baseline – Less preferred compartment made CS+ • Conditioning – Experimental group: CS+ paired with sexually receptive female (US) – Control group: US only in home cage, never in CS+ compartment
% time spent in CS+ compartment Results Experimental Control 1 2 Preference Tests 3
Conditional Relations • Binary relations: between two events (CS-US, operant-outcome) • Modulator: a third event that determines the nature of a binary relation • Modulator signals a conditional relation
Modulators • Instrumental – S+: respond --> reinforcer – S-: respond --> no reinforcer • Classical – “Facilitators” or “occasion setters”, not excitatory or inhibitory conditional stimuli – CS = noise, US = food, modulator = light noise food
- Slides: 50