Cognitive Scientist Explains Explanatory Coherence and Computational Philosophy

00:05:12
https://www.youtube.com/watch?v=SyCyNvCMu68

Sintesi

TLDRExplanatory coherence is the principle that the best explanation is one which is most consistent with the evidence available. This concept is illustrated through Darwin’s work in 'The Origin of Species', where he presented a coherent argument for evolution via natural selection against divine creation. The speaker connects this theory to the computational modeling of coherence using neural networks, positing that the brain functions as a coherence engine—constantly making sense of past experiences and future predictions by satisfying various constraints. This has profound implications for psychology and AI, supporting the idea that coherence is essential in decision-making and human cognition.

Punti di forza

  • 🌍 Explanatory coherence helps explain phenomena, like evolution.
  • 📖 Darwin's 'Origin of Species' exemplifies coherence in theory.
  • 🤖 Neural networks can model coherence computationally.
  • 🧠 The brain acts as a coherence engine for sense-making.
  • 💡 Inference to the best explanation aids in scientific reasoning.
  • 📊 Coherence is crucial for AI in mimicking human decision-making.
  • 🔗 Coherence connects past experiences with future predictions.
  • 🎓 Coherence fosters deeper understanding in psychology.
  • 🛠️ Decision-making relies on evaluating multiple constraints.
  • ✝️ Coherence contrasts with divine creation as an explanation.

Linea temporale

  • 00:00:00 - 00:05:12

    The speaker discusses explanatory coherence, illustrating it through Darwin's work in 'The Origin of Species', where Darwin provides a coherent explanation for his observations about species evolution, challenging the prevailing belief in divine creation. The speaker explains that the coherence of an argument, such as Darwin's natural selection, can be evaluated through philosophical inference to the best explanation. They highlight their research background, particularly in neural networks, which connects computational theory with psychological understanding of coherence, positing that the mind operates as a coherence agent, integrating various constraints to form a coherent understanding of the world, suggesting significant implications for AI, psychology, and cognitive science.

Mappa mentale

Video Domande e Risposte

  • What is explanatory coherence?

    Explanatory coherence is the idea that a theory or explanation is more valid if it is consistent and aligns well with the evidence available.

  • How did Darwin contribute to the concept of explanatory coherence?

    Darwin provided a coherent explanation for species evolution, favoring natural selection over divine creation, creating an argument that was scientifically supported.

  • What role do neural networks play in understanding coherence?

    Neural networks can computationally model and explain the concept of coherence by evaluating and satisfying a set of constraints.

  • How does this concept relate to psychology?

    Explanatory coherence helps in understanding human cognition as the brain operates to create coherent explanations and predictions about the world.

  • What are the implications of coherence in AI?

    In AI, coherence is crucial for algorithms to effectively simulate human-like decision-making and understanding.

  • Can coherence be quantified?

    Yes, coherence can be represented computationally, allowing for precise evaluations within cognitive science.

  • What is a coherence engine?

    A coherence engine is a metaphor describing how the brain organizes information and experiences to form coherent narratives or predictions.

  • How does coherence impact decision-making?

    Coherence affects decision-making by forcing individuals to weigh different constraints and select the most logical or evidence-supported option.

  • What is 'inference to the best explanation'?

    Inference to the best explanation is a philosophical approach that suggests we choose the explanation that best fits the available evidence.

  • Why is coherence considered fundamental in cognitive science?

    Coherence allows for a deeper understanding of how we process information and make sense of our experiences in everyday life.

Visualizza altre sintesi video

Ottenete l'accesso immediato ai riassunti gratuiti dei video di YouTube grazie all'intelligenza artificiale!
Sottotitoli
en
Scorrimento automatico:
  • 00:00:00
    I want to discuss uh explanatory
  • 00:00:03
    coherence uh you've done a lot of work
  • 00:00:05
    on that and I'm wondering if you could
  • 00:00:06
    kind of give a brief introduction to
  • 00:00:09
    what uh explanatory coherence is to a
  • 00:00:12
    lay audience while at the same time uh
  • 00:00:15
    why you think uh it's what what are the
  • 00:00:18
    what are the implications in fields such
  • 00:00:20
    as Ai and
  • 00:00:21
    psychology oh that's a big question but
  • 00:00:23
    it's a really interesting one so let's
  • 00:00:25
    go back to Darwin so I mentioned one of
  • 00:00:27
    my favorite books of all time is the
  • 00:00:29
    Origin of Species
  • 00:00:30
    and so what was he doing in that uh well
  • 00:00:33
    in my view what he was trying to do was
  • 00:00:34
    to give a coherent explanation of all
  • 00:00:36
    sorts of things that he observed so he
  • 00:00:38
    went on this Voyage around the world and
  • 00:00:40
    he collected all sorts of other kinds of
  • 00:00:41
    biological information and it gradually
  • 00:00:44
    seemed to him that it seemed in fact
  • 00:00:47
    that the species had evolved I mean now
  • 00:00:49
    everybody knows that kids get that
  • 00:00:51
    probably in grade four but it was a a a
  • 00:00:53
    very controversial idea because some
  • 00:00:55
    people had maintained it but it went up
  • 00:00:57
    against religious doctrines and so he
  • 00:01:00
    gradually started to amass more and more
  • 00:01:02
    evidence that species had evolved but
  • 00:01:06
    then from reading a crazy Economist
  • 00:01:09
    named malus he suddenly got the idea of
  • 00:01:10
    how they evolved and that's how he came
  • 00:01:12
    up with the idea of of of natural
  • 00:01:14
    selection so now we had not only a bunch
  • 00:01:16
    of observations and the idea that
  • 00:01:19
    Evolution probably had occurred that
  • 00:01:20
    would explain it but an idea of how how
  • 00:01:23
    Evolution had occurred that is that
  • 00:01:25
    natural selection was the mechanism
  • 00:01:27
    behind Evolution so when he did in that
  • 00:01:30
    book was an incredibly beautiful
  • 00:01:32
    argument for his view as opposed to the
  • 00:01:35
    view that was dominant at the time which
  • 00:01:37
    was divine creation so what he was
  • 00:01:40
    trying to show is that his view was a
  • 00:01:43
    better explanation but than Divine
  • 00:01:46
    creation because it was more coherent
  • 00:01:48
    with the evidence um so this is an
  • 00:01:51
    account that philosophers call inference
  • 00:01:53
    to the best explanation you can argue
  • 00:01:56
    that something's the best explanation
  • 00:01:57
    because it's more coherent with the
  • 00:01:59
    evidence but now you have to say what
  • 00:02:00
    coherence is so I had these early ideas
  • 00:02:03
    coming out of my philosophy of science
  • 00:02:05
    background but then in
  • 00:02:09
    1987 I got one of the best ideas I've
  • 00:02:12
    ever had which was how to turn coherence
  • 00:02:14
    from a sort of vague philosophical idea
  • 00:02:16
    into a precise computational one that is
  • 00:02:19
    how can you compute coherence so I've
  • 00:02:22
    been working on neural networks in
  • 00:02:24
    collaboration with my uh uh colleague
  • 00:02:28
    Keith Holio and he came came up with an
  • 00:02:30
    idea that you could use neural networks
  • 00:02:31
    to explain analogy uh these neural
  • 00:02:34
    networks of course are now absolutely
  • 00:02:36
    Central to artificial intelligence it's
  • 00:02:38
    it's really taken off that's a whole
  • 00:02:40
    fascinating Topic in itself um but he
  • 00:02:43
    figured out a way of doing that and by
  • 00:02:45
    that time I'd done my masters in
  • 00:02:46
    computer science so I was a pretty good
  • 00:02:47
    programmer and so I programmed up a
  • 00:02:50
    program to use neural networks to do
  • 00:02:52
    analogies um so that that was nice and
  • 00:02:55
    then I thought what else could apply to
  • 00:02:57
    and then I thought back to the problem
  • 00:02:58
    that was part of my doctoral
  • 00:02:59
    dissertation which was inference to the
  • 00:03:01
    best explanation how do you pick up a
  • 00:03:03
    best theory and so then I realized that
  • 00:03:05
    that kind of coherence can be understood
  • 00:03:08
    using the same kind of neural network
  • 00:03:10
    technique that Keith and I had done for
  • 00:03:12
    for analogy so it was a really powerful
  • 00:03:15
    method both computationally but also
  • 00:03:18
    psychologically because there now since
  • 00:03:19
    then been lots of psychological
  • 00:03:21
    experiments that back up this idea of
  • 00:03:23
    coherence so I think of the Mind the
  • 00:03:25
    brain as essentially a coherence agent
  • 00:03:27
    there's some people who think that it's
  • 00:03:29
    primarily a predictive engine but I
  • 00:03:30
    don't think that's true I think it's
  • 00:03:32
    primarily a a coherence engine we're
  • 00:03:34
    trying to make sense of things whether
  • 00:03:35
    we're making sense of the past which is
  • 00:03:38
    what explanations do or making sense of
  • 00:03:40
    the future we're making coherent
  • 00:03:42
    predictions or we're trying to identify
  • 00:03:44
    things is is that a rabbit or a squirrel
  • 00:03:47
    those are are different kinds of so
  • 00:03:49
    everything we do can be understood as
  • 00:03:51
    having coherence behind it but coherence
  • 00:03:54
    Now isn't just a sort of vague metaphor
  • 00:03:56
    that it was for philosophers it's not
  • 00:03:58
    just a matter of consistency it's rather
  • 00:04:00
    of taking a whole bunch of different
  • 00:04:02
    things and putting them into a good
  • 00:04:03
    package but what's a good package well
  • 00:04:06
    here there's a an idea that came out of
  • 00:04:09
    the um neural network world called
  • 00:04:11
    constraint satisfaction so we're trying
  • 00:04:13
    to satisfy a bunch of constraints what
  • 00:04:16
    constraints did Darwin face while he was
  • 00:04:18
    trying to explain as much as possible
  • 00:04:20
    about what he'd seen in the biological
  • 00:04:22
    world that's a positive constraints but
  • 00:04:25
    he also had a negative constraint is
  • 00:04:26
    that he had to show that he could do
  • 00:04:28
    that better than the theory that was the
  • 00:04:30
    computer at the time competitor at the
  • 00:04:32
    time which is divine creation so that's
  • 00:04:33
    a negative constraint so what you're
  • 00:04:35
    doing in all of these things whether's
  • 00:04:37
    decision- making or pattern recognition
  • 00:04:40
    or even emotion you're putting together
  • 00:04:43
    different sorts of constraints to
  • 00:04:44
    evaluate what's the most coherent view
  • 00:04:47
    so that's how I came to see coherence
  • 00:04:49
    not just as a vague philosophical idea
  • 00:04:51
    but as a quite precise computational one
  • 00:04:54
    that can be used to explain the
  • 00:04:57
    mechanisms that underly a vast amount of
  • 00:04:59
    human things
  • 00:05:00
    so that's why I think coherence is
  • 00:05:02
    really a fundamental idea to psychology
  • 00:05:05
    and cognitive science and to into these
  • 00:05:09
    philosophical projects as well
Tag
  • explanatory coherence
  • Darwin
  • natural selection
  • neural networks
  • cognitive science
  • philosophy
  • AI
  • psychology
  • decision-making
  • constraints