1 / 40

A naturalistic approach to perception and action Sandro Nannini (University of Siena)

A naturalistic approach to perception and action Sandro Nannini (University of Siena). Naturalising mental states. A mental State X is naturalised iff:

tana
Download Presentation

A naturalistic approach to perception and action Sandro Nannini (University of Siena)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. A naturalistic approach to perception and actionSandro Nannini(University of Siena) Hannover, 9th June 2007

  2. Naturalising mental states • A mental State X is naturalised iff: • We know how X is implemented by a functional state Y that has the same causes and brings about the same effects than X (functional reduction). • We know how the functional state Y is implemented by a brain process Z (neurological implementation). “X” belongs to the language of folk psychology, “Y” to the language of cognitive psychology, “Z” to the language of neurosciences. Hannover, 9th June 2007

  3. Naturalising mental states • Functional reduction [1] is a task of cognitive psychologists, neurological implementation [2] is a task of neuroscientists. • The step (2) presupposes the step (1) and vice versa (top down explanation and bottom up explanation). • Both steps presuppose a previous conceptual analysis and modification of the language of folk psychology (this is the task of philosophers). Hannover, 9th June 2007

  4. Perception and sensory-motor coordination • Animals acquired the ability to perceive some features of the external world and of their own body in order to execute movements apt to increase the probability to survive (e.g. by catching preys or avoiding plunderers). Hannover, 9th June 2007

  5. Perception and sensory-motor coordination • Human senses and human sensori-motor coordination are the result of biological evolution. Biological evolution Hannover, 9th June 2007

  6. Perception and sensory-motor coordination:Representation-Action Theory (RAT) • Perceptions can be conscious or unconscious: in both cases they are mental representations of the internal and external world. • Human beings construct a representation of the external world in order to move and act in it. Hannover, 9th June 2007

  7. The computational brain • According to the RAT the brain acquires by means of the senses a certain amount of information about some regularities of the external world as regards the distribution of matter and physical events in space and time and changes the format of such information step by step until a pattern of motor neurons activity able to trigger a right motor response is produced. Hannover, 9th June 2007

  8. Styles of brain computation No!!! Symbolic representations A Brooks’ robot Unlikely! No representations May be! Subsymbolic representations Hannover, 9th June 2007

  9. Naturalising perceptions according to the RAT Functional reduction • A perception is functionally reducible to an intermediate step in the information processing of sensori-motor coordination • Therefore it is similar to the activity pattern of hidden units in an artificial neural network and is describable as a vector in a state space. Hannover, 9th June 2007

  10. Naturalising perceptions according to the RAT Neural implementation • Perceptions as vectors in a space state are biologically implemented by the dynamics of brain processes. Hannover, 9th June 2007

  11. Criticisms on the RAT • The sensory motor coordination does not need any inner representation of the external world (Antirepresentationalism; e.g. Brooks’ robots). • Perceptions are conscious. Brain processes cannot be conscious (Mind-Body Problem: can consciousness be naturalised?). • Perceptions are ‘intentional’ states. No brain process can be ‘intentional’ (semantic relations are not reducible to causal relations): • The possibility of deceptive perceptions • The neural implementation of contents • The identification of a real object by means of a phenomenal object. Hannover, 9th June 2007

  12. Criticisms on the RAT • The sensory motor coordination does not need any inner representation of the external world (Antirepresentationalism; e.g. Brooks’ robots). Hannover, 9th June 2007

  13. A reply to (1) Mice in a labyrinth Hannover, 9th June 2007

  14. A reply to (1) • M1s recognise red (R), green (G) and yellow (Y) colours and follow these rules: • ‘If R go to right’ • ‘If G go to left’, • ‘If Y go ahead’ • Therefore they are able to reach the food if they enter the labyrinth from South. If they enter from West they have no chance to reach the food. Hannover, 9th June 2007

  15. A reply to (1) M2s instead enter the labyrinth sometimes from South and sometimes from West. They can reach the food in both cases because they are able to remember from which entrance they came and to learn by trial and error that if they enter from South they must follow the rule (1) but if they enter from West they must follow the rule (1*), that is, ‘If R go ahead’. Hannover, 9th June 2007

  16. A reply to (1) M3s are able to construct by trials and errors a map of the labyrinth. They remember that after entering the labyrinth from South they went to East as they turned to left at the red square. Moreover they somehow have learned (or have the inborn knowledge) that East and West are opposite directions. Therefore, even if they never used the entrance West they are able to forecast that if they entered the labyrinth from West they should go ahead at the red square instead of going to right as they always did till that moment in their life. Hannover, 9th June 2007

  17. A reply to (1) In other words, the richer and more ‘objective’ the representation of the labyrinth is the more flexible and efficient the behaviour of the mouse is. Similarly human beings cannot work like Brooks’ robots. They need an inner representation of the environment in which they act. Otherwise their behaviour would not be so flexible as it is. Mental representations are in humans ‘multi-purpose’ representations. Hannover, 9th June 2007

  18. Criticisms on the RAT • Perceptions are conscious. Brain processes cannot be conscious (Mind-Body Problem: can consciousness be naturalised?). Hannover, 9th June 2007

  19. A reply to (2) There is a double disjunction between perception and consciousness: • Blindsight • Anton’s syndrome Hannover, 9th June 2007

  20. Blindsight Perception without consciousness Blindsight allows people to use visual information they get through their eyes even though they have no consciousness of the visual experience. Hannover, 9th June 2007

  21. Anton’s syndrome Consciousness without perception The doctor asks what is on the table in front of you: “A clock” you say as though the doctor is an idiot. Hannover, 9th June 2007

  22. A reply to (2) • Conclusion: you can perceive something without been aware of perceiving it and you can be aware of perceiving something without really perceiving it. • Consciousness and perception are two distinct phenomana implemented by different brain processes. • The RAT naturalises only ‘mere perceptions’, that is, the functional basis common to all perceptions independently of their being conscious or unconscious. Hannover, 9th June 2007

  23. Criticisms on the RAT • Perceptions are ‘intentional’ states. No brain process can be ‘intentional’ (semantic relations are not reducible to causal relations): • The possibility of deceptive perceptions Hannover, 9th June 2007

  24. A reply to (3a) R F. Dretske’s theory: R(O*) is true iff OB is true. A girl sees a cat and her perception is true because it is caused by an animal that is really a cat. O* B B O S Hannover, 9th June 2007

  25. An objection on Dretske‘s theory R The girl is looking at a squirrel but in the dark she sees a cat: O’ R(O*) is true but R(O*) is false. O* B B O‘ S Hannover, 9th June 2007

  26. The RAT reply to (3a) The girlspursuesthe goal G to carry the catin her arms (D(G*)) and executes an action A in order to get G. Perceiving the cat (R(O*)) is necessary to get G because A is caused by D(G*) & R(O*): OR(O*) [& D(G*)]AG Hannover, 9th June 2007

  27. A reply to (3a) D(G*) R(O*) B G A O T1 T2 Hannover, 9th June 2007

  28. A reply to (3a) The RAT Dretske’s theory Mind R(W* S*) Mind R(W*) R(S*) R(O*) Emotions R(O*) D(G*) Body Body B B Hand etc. S S G Physical world O Physical world O Hannover, 9th June 2007

  29. A reply to (3a) (a) is a valid objection against Dretske’s theory but not against the RAT. Hannover, 9th June 2007

  30. Criticisms on the RAT • Perceptions are ‘intentional’ states. No brain process can be ‘intentional’ (semantic relations are not reducible to causal relations): • Semantic relations are necessary conceptual relations. Causal relations are contingent empirical relations: how can a chain of causal relations implement a semantic relation? Hannover, 9th June 2007

  31. A reply to (3b) Contents can be naturalised by means of the ‘adverbial theory’ (of perceptions etc.). “I see red” means “I see redly”: the content of a mental representation is part of its form. Therefore it can be implemented by certain physical properties of a brain process. Hannover, 9th June 2007

  32. An objection to the adverbial theory 1 1 I see a green square I see a red circle Distinguishing between (1) and (2) is impossible. I see greenly, squarely, redly, and circulary I see a green circle I see a red square 2 2 Hannover, 9th June 2007

  33. A reply to 3b(In defence of adverbial theory) • Many neuroscientists think that the form and the colour of a same object are connected by the brain thanks to a common temporal code (synchronisation). • Therefore the difference between perceiving (1) and perceiving (2) might be implemented by the difference between two dynamics of the brain. Hannover, 9th June 2007

  34. Criticisms on the RAT • Perceptions are ‘intentional’ states. No brain process can be ‘intentional’ (semantic relations are not reducible to causal relations): • A functional state of the brain cannot identify the real object from which is caused and on which the action of the agent is directed. Therefore the RAT is insufficient to explain why perceptions are instead sufficient to identify external objects. Hannover, 9th June 2007

  35. A reply to 3c: the 1-eaters and the 2-eaters 1000200000010000000200100000000001222200000010000000200100000000001222?????????????????????????????? 100020000001000 000020010000000 000122220000001 000000020010000 0000001222????? ??????????????? ?????????? 1000200000 0100000002 0010000000 0001222200 0000100000 0020010000 0000001222 ?????????? ?????????? ?????????? Which is the right representation? It depends on what you eat! Hannover, 9th June 2007

  36. A reply to (3c): frogs and flies • A frog recognizes flies as food only if they are moving. • We human beings instead recognize flies as flies independently of their movements. • Therefore, the representation that an animal has of its environment is functional to the actions that it is able to execute in that environment. Hannover, 9th June 2007

  37. A reply to (3c): frogs and flies • It is not the case that we human beings see flies as they are, frogs instead see them as they appear to them. Hannover, 9th June 2007

  38. A reply to (3c) • Mental representations are constructions of the mind (=brain), not the copies of real objects. • There is similarity between the activity patterns of hidden units in an artificial neural network and mental representations: they are a representation (= a state space partition) of the input apt to get the desired output. • Every species lives in its own phenomenal world adapted to a certain kind of interaction with the real physical world. Hannover, 9th June 2007

  39. Conclusion • Perceptions can be naturalised only if the common sense concept of ‘perception’ is radically changed: perceptions are not copies of real objects passively received from the external world but formats given to sensory inputs in order to construct a stable and multi-purpose model of reality that is able to control the very flexible behaviour of human beings. • These perceptions can be functionally reduced and therefore can be implemented by brain processes. Hannover, 9th June 2007

  40. Thank you for your attention! Hannover, 9th June 2007

More Related