Science, Technology & the Future | Stevan Harnad - Symbol Grounding part 2 @scfu | Uploaded December 2020 | Updated October 2024, 1 hour ago.
Further discussion with Stevan Harnad on Symbol Grounding - including discussion on Extended Cognition, consciousness, meaning, hive minds, GPT-3, a-life, grounding categories, vanishing intersections, natural kinds, essences & transitions, evolutionary psychology, learning & skinner boxes, dictionaries & grounding sets, shared cognition, the identity of indiscernibles, ontology and feeling primitives.
02:33 extended cognition vs extended consciousness
03:57 Does every part of our modular brain contribute to consciousness?
05:04 Disagreement on the meaning of the word 'cognition'
07:25 Wittgenstein - the meaning of a word is it's use
08:51 Shared grounding in a hive?
10:40 Extending reach vs extending hands or arms
12:34 GPT3 as a writing aid. Chatting with dead philosophers
16:30 Lost in the hermeneutic hall of mirrors
19:59 A metaphor for symbol grounding?
21:54 Further on what symbol grounding is.. categories from propositions - affording the capacity to acquire and transmit grounded categories through language (instead of by trial and error)
25:34 Artificial life - mushroom (toy) example of symbol grounding
32:30 Grounding categories are not just concrete objects, but can be abstract concepts (i.e. sharpness)
33:42 GPT3 Goethe revisited.
37:22 Vanishing intersections. Chompsky's work on universal grammar (syntax) & the poverty of stimulus.
48:34 How do humans learn categories? 3 ways: supervised, unsupervised and through language
52:28 What happens when we find evidence to dis-confirm something we have already learned? (i.e. a black swan)
55:46 Natural kinds
59:05 Natural kinds by transition?
02:15 Essences and transitions
05:48 How much of our current behavior have homolouges in the ancestral environment?
10 Selective behavior - example of the peacock's tail that controls for cheating
12:15 Paraciticm
14:13 To what degree did our capacity for generalization come before grammar?
14:48 Learning, Skinner boxes, abstraction and categorization. Is (category) learning itself is the mother of all generalizations?
18:58 Dictionaries & Grounding sets
19:53 What is meant by a dictionary? The nucleus (can define all words inside and outside itself), the core (can define all inside itself), satellites (around the core, tiny clusters of words). Minimal grounding sets are not dictionaries - they can only define words outside of itself, it is also the smallest amount of words which can define all the rest of the words by combinations of words - turns out there are many of them - so finding the ultimate grounding set is an N-P complete problem (using a directed graph). All grounding sets are part core and part satellite. The size of the minimum grounding set is between 750 to 1500 words.
30:48 Grounding sets, and the dictionary game.
35:03 The easy and the hard problems of consciousness. Why is the hard problem of consciousness hard?
39:56 Shared cognition? Shared experience? The case of the Hogan sisters, Siamese twins conjoined at the head - sharing parts of their brains.
44:21 The identity of indiscernibles
48:09 Renounce ontology! Psychologists Stevan Harnad is a naive realists, he is interested in what organisms can do, and how they feel.
49:28 Feeling primitives - can feels be reduced to monads or are they complex?
Many thanks for tuning in!
Consider supporting SciFuture by:
a) Subscribing to the SciFuture YouTube channel: youtube.com/subscription_center?add_user=TheRationalFuture
b) Donating
- Bitcoin: 1BxusYmpynJsH4i8681aBuw9ZTxbKoUi22
- Ethereum: 0xd46a6e88c4fe179d04464caf42626d0c9cab1c6b
- Patreon: patreon.com/scifuture
c) Sharing the media SciFuture creates
Kind regards,
Adam Ford
- Science, Technology & the Future - #SciFuture - scifuture.org
Further discussion with Stevan Harnad on Symbol Grounding - including discussion on Extended Cognition, consciousness, meaning, hive minds, GPT-3, a-life, grounding categories, vanishing intersections, natural kinds, essences & transitions, evolutionary psychology, learning & skinner boxes, dictionaries & grounding sets, shared cognition, the identity of indiscernibles, ontology and feeling primitives.
02:33 extended cognition vs extended consciousness
03:57 Does every part of our modular brain contribute to consciousness?
05:04 Disagreement on the meaning of the word 'cognition'
07:25 Wittgenstein - the meaning of a word is it's use
08:51 Shared grounding in a hive?
10:40 Extending reach vs extending hands or arms
12:34 GPT3 as a writing aid. Chatting with dead philosophers
16:30 Lost in the hermeneutic hall of mirrors
19:59 A metaphor for symbol grounding?
21:54 Further on what symbol grounding is.. categories from propositions - affording the capacity to acquire and transmit grounded categories through language (instead of by trial and error)
25:34 Artificial life - mushroom (toy) example of symbol grounding
32:30 Grounding categories are not just concrete objects, but can be abstract concepts (i.e. sharpness)
33:42 GPT3 Goethe revisited.
37:22 Vanishing intersections. Chompsky's work on universal grammar (syntax) & the poverty of stimulus.
48:34 How do humans learn categories? 3 ways: supervised, unsupervised and through language
52:28 What happens when we find evidence to dis-confirm something we have already learned? (i.e. a black swan)
55:46 Natural kinds
59:05 Natural kinds by transition?
02:15 Essences and transitions
05:48 How much of our current behavior have homolouges in the ancestral environment?
10 Selective behavior - example of the peacock's tail that controls for cheating
12:15 Paraciticm
14:13 To what degree did our capacity for generalization come before grammar?
14:48 Learning, Skinner boxes, abstraction and categorization. Is (category) learning itself is the mother of all generalizations?
18:58 Dictionaries & Grounding sets
19:53 What is meant by a dictionary? The nucleus (can define all words inside and outside itself), the core (can define all inside itself), satellites (around the core, tiny clusters of words). Minimal grounding sets are not dictionaries - they can only define words outside of itself, it is also the smallest amount of words which can define all the rest of the words by combinations of words - turns out there are many of them - so finding the ultimate grounding set is an N-P complete problem (using a directed graph). All grounding sets are part core and part satellite. The size of the minimum grounding set is between 750 to 1500 words.
30:48 Grounding sets, and the dictionary game.
35:03 The easy and the hard problems of consciousness. Why is the hard problem of consciousness hard?
39:56 Shared cognition? Shared experience? The case of the Hogan sisters, Siamese twins conjoined at the head - sharing parts of their brains.
44:21 The identity of indiscernibles
48:09 Renounce ontology! Psychologists Stevan Harnad is a naive realists, he is interested in what organisms can do, and how they feel.
49:28 Feeling primitives - can feels be reduced to monads or are they complex?
Many thanks for tuning in!
Consider supporting SciFuture by:
a) Subscribing to the SciFuture YouTube channel: youtube.com/subscription_center?add_user=TheRationalFuture
b) Donating
- Bitcoin: 1BxusYmpynJsH4i8681aBuw9ZTxbKoUi22
- Ethereum: 0xd46a6e88c4fe179d04464caf42626d0c9cab1c6b
- Patreon: patreon.com/scifuture
c) Sharing the media SciFuture creates
Kind regards,
Adam Ford
- Science, Technology & the Future - #SciFuture - scifuture.org
