Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Animal Learning theory lecture

81 views

Published on

Presentation on basic learning theory for teaching Animal Behaviour.

This presentation is developed by Ralph Thompson and based on one used at the University of Edinburgh with elements (particularly images) removed for copyright reasons.
It is intended as a basis for lectures, to which images/graphics can be added and the lecture adapted for the purposes needed. The use of a coloured background to aid accessibility may benefit some students.

Reflective/class questions are shown in light blue. Addition of other reflective or class response questions to check understanding and promote engagement is recommended.

Published in: Science
  • Be the first to comment

Animal Learning theory lecture

  1. 1. Dr Ralph R.J. Thompson BA (Cantab.) MSc (Oxon.) PhD (Bris.) FHEA University of Edinburgh © Ralph Thompson 2018, University of Edinburgh Licensed under a Creative Commons Attribution licence 4.0 All images CC0, Pixabay Animal Learning How animals change their behaviour in response to information from their environment
  2. 2. This presentation is developed by Ralph Thompson and based on one used at the University of Edinburgh with elements (particularly images) removed for copyright reasons. It is intended as a basis for lectures, to which images/graphics can be added and the lecture adapted for the purposes needed. The use of a coloured background to aid accessibility may benefit some students. Reflective/class questions are shown in light blue. Addition of other reflective or class response questions to check understanding and promote engagement is recommended. © Ralph Thompson 2018, University of Edinburgh Licensed under a Creative Commons Attribution licence 4.0 NOTES ON THIS PRESENTATION by Dr Ralph Thompson University of Edinburgh
  3. 3. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Objectives • To introduce learning theory to you if you are new to the ideas • To make you think about learning in some new ways if you already know learning theory (not all Pavlov and Skinner) – Learning in the context of ethology/emotion • Giving you the knowledge to… – understand why animals change their behaviour with experience – plan the training of animals, but more practical training later on (including in your assessment) – Understand and critically evaluate training methods used by others
  4. 4. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Overview • What is Learning? – stimuli • Types of learning – Non-associative – Associative • Reinforcement – Shaping – Factors affecting reinforcement • Examples from research Image CCO, Pixabay
  5. 5. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 What is Learning? A (relatively long-lasting) change in behaviour as the result of experience • This can be any experience and any behaviour • An adaptive process that allows an animal to match it’s behaviour to the environment that it find itself in • If the experience is intentionally provided for an animal by humans the learning process may be referred to as training (but animals will learn from human stimuli in unintended ways as well)
  6. 6. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Learning… • ...is a form of cognition • …involves/motivated by emotion
  7. 7. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Learning in understanding behaviour • Tinbergen’s 4 whys: o Adaptation o Phylogeny o Mechanism o Ontogeny – Cognition (and physiology and anatomy) – Learning (and physical development) Behaviourism and cognitivism
  8. 8. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Doing isn’t always understanding “though the practice of a science is known to all, only a few persons are acquainted with the rules and laws on which the science is based. …riders of horses and elephants train these animals without knowing the science of training animals, but from practice only.” The Kama Sutra of Vātsyāyana (3rd century AD, Richard Burton Translation) Image CCO, Pixabay
  9. 9. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Why learn? Learning allows animals to change their behaviour in response to experience • Behaviour can fit novel or changing environments • Behaviours can be formed that are too complex to encode genetically Learning occurs all the time • In the wild • In captivity • During training
  10. 10. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Stimuli • We call quanta of information that are sensed by the animal and cause a response stimuli (singular stimulus) – The response can be immediate behaviour or learning or both – We are most often concerned with external (environmental) stimuli but they can also be internal to the animal (e.g. internal pain) – There are infinite number of potential stimuli in nature…
  11. 11. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Stimulus importance • Animals needs to know which stimuli to respond to. – They must have a way of identifying which environmental information is most important to them – How can they do this? • ‘Innate’ value / salience – What kind of stimuli should be most salient for animals? – What stimuli are most salient for you? – Are innate evolved responses enough? When would they be most/least useful?
  12. 12. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Stimulus importance • Learned value – Determining which stimuli are meaningful through repeated exposure or exposure during sensitive periods (non-associative learning) – Pairing meaningful stimuli that are related to each other (associative learning) • Both are constantly updated during the life of an animal, though the animal may be more sensitive at particular periods • May also depend on motivational state (e.g. hunger)
  13. 13. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Associate and non-associative learning • Non-associative – Imprinting – Habituation – Sensitisation & desensitisation • Associative – Classical Conditioning (Pavlovian) – Operant Conditioning (Skinnerian)
  14. 14. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Non-associative learning Add image
  15. 15. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Imprinting Relatively permanent, rapid, learning during a particular sensitive period Image CCO, Pixabay
  16. 16. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Habituation A decrease (or cessation) in response following repeated exposure to the stimulus • Stimulus is not (or is not longer) biologically relevant • Can be generalised (apply to similar stimuli) • Particularly important in prey animals (as they are likely to be neophobic) • To be effective must be gradual and not forceful • May occur to accidental/unintentional stimuli (e.g. riding school horses)
  17. 17. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Sensitisation • Generally aversive stimuli (e.g. pain, loud noises) An increase in response following repeated exposure to the stimulus Image CCO, Pixabay
  18. 18. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Desensitisation • Similar to habituation, but to a previously sensitised stimulus A decrease (or cessation) of a previously sensitised response following repeated exposure to the stimulus
  19. 19. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Associative learning Add image
  20. 20. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Classical Conditioning • Also known as Pavlovian Conditioning • Pairing of a conditioned stimulus with an unconditioned stimulus • The CS must predict the US • Produces a conditioned response (CR) to the CS similar to the unconditioned response (UR) to the US • Associates with novel stimuli with appropriate responses • e.g. salivating in response to sound of food being prepared A learned association between a previously neutral stimulus (CS) and a previously meaningful stimulus (US)
  21. 21. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Conditioned responses Bell (CS) Food (US) Food (US) Salivation (UR) Pairing Unconditioned Response (CR)
  22. 22. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Classically conditioned Classical conditioning will be happing all the time, to you animals and to you, without you knowing it. This can be helpful but can also cause problems – Classical conditioning is used to establish the association between the click and the reward in clicker training – Classical conditioning can make an animal show a fearful response to a vet if the vet previously performed a painful procedure (e.g. injection)
  23. 23. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Operant Conditioning • Also known as Skinnerian conditioning • The animal learns that a behaviour is followed by a consequence that it likes (a reinforcer) or does not like (a punisher) and behaves accordingly in the future • The foundation of much of animal training A change in the probability of a behaviour being performed due to a learned association between that behaviour and a meaningful consequence for the animal
  24. 24. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Reinforcers and punishers • Reinforcement and punishment • Primary and secondary reinforcers • Reward prediction error • Reinforcement schedules – Fixed – Intermittent (fixed/variable) • Extinction • Examples
  25. 25. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Reinforcers and emotion • Reinforcement and punishment referring to change in the affective state of the animal and hence the change in likelihood of performing a behaviour which it is contingent upon • Reinforcers and punishes can be appetitive and aversive…
  26. 26. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Reinforcers and punishers table Add (+) a stimulus Subtract (-) a stimulus Increase (reinforce) the behaviour Positive Reinforcement Negative Reinforcement Decrease (punish) the behaviour Positive Punishment Negative Punishment
  27. 27. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Primary Reinforcers • Have an evolved biological meaning for the animal • Things that should be approached or avoided in the ancestral environment – Not necessarily still useful
  28. 28. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Secondary Reinforcers • Have become reinforcing / punishing through learned association (classical conditioning) – e.g. clicker Add image
  29. 29. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Reinforcement schedules • Continual reinforcement – every occurrence of the CS/behaviour is paired with the US/reinforcer • Partial reinforcement – only some of the some occurrences of the CS/behaviour are paired with the US/reinforcer – The lower the proportion of CS/behaviours that are paired with the US/reinforcer the more slowly conditioning occurs – Can be interval or ratio, fixed or variable
  30. 30. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Reinforcement • When the probability of the US in the absence of the CS increases the strength of learning decreases – The CS must be a useful predictor of the US • The CS can also predict the absence of a US (reduced prob. of US following CS) – This is inhibitory conditioning
  31. 31. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Extinction Once pairing has occurred the response can be extinguished if the CS is presented to the animals without the US, or the behaviour is performed with the reinforcer/punisher – The strength of the CR declines gradually – CR acquired through partial reinforcement takes longer to extinguish (because the animal is not expecting all occurrences to be paired)
  32. 32. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Equipotentiality? • Equipotentiality – any animal can be conditioned to perform any response with any stimulus as long as it can perform the behaviour and sense the stimuli • Is this what we observe? • Some behaviours may seem naturally more appropriate to some contingencies (e.g. escape behaviours for an electric shock)
  33. 33. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Factors affecting learning • Motivational state – e.g. hunger • Biological predisposition • Environment – Distractors – Anxiety • Age – e.g. horses handled when young show lower neophobia – e.g. old dogs decline in learning and memory ability
  34. 34. © Ralph Thompson 2018, University of Edinburgh, licensed under a Creative Commons Attribution licence 4.0 Summary • Learning changes the behaviour of an animal in response to it’s experience and occurs all of the time • Learning can be non-associative (as with habituation and sensitisation) • …or associative (as with classical and operant conditioning) • Reinforcement and punishment change the probability of an animal performing an operant behaviour • These can be used to train animals though shaping • Timing and strength of predictive links between CS and US are important • It is important to recognise the effects of motivation and sensory bias when training or understanding learning
  35. 35. This presentation is developed by Ralph Thompson and based on one used at the University of Edinburgh with elements (particularly images) removed for copyright reasons. It is intended as a basis for lectures, to which images/graphics can be added and the lecture adapted for the purposes needed. The use of a coloured background to aid accessibility may benefit some students. Reflective/class questions are shown in light blue. Addition of other reflective or class response questions to check understanding and promote engagement is recommended. © Ralph Thompson 2018, University of Edinburgh Licensed under a Creative Commons Attribution licence 4.0 NOTES ON THIS PRESENTATION by Dr Ralph Thompson University of Edinburgh

×