predictive processing
Article
predictive processing is a recurring concept in the Astral Codex Ten archive, appearing 3 times across 3 issues between January 21, 2021 and July 16, 2024. The archive places it in contexts such as “the predictive processing community is starting to really chip away at the question of why psychotherapies work”; “Link to Predictive Processing”; “my introspection about my information processing makes more sense since I learned about predictive processing”. It most often appears alongside Eliezer Yudkowsky, fMRI, Neuralink.
Metadata
- Category: Concepts
- Mention count: 3
- Issue count: 3
- First seen: January 21, 2021
- Last seen: July 16, 2024
Appears In
- Still Alive
- Your Book Review: Are We Smart Enough To Know How Smart Animals Are?
- Consciousness As Recursive Reflections
Related Pages
-
- Eliezer Yudkowsky (2 shared issues)
-
- fMRI (2 shared issues)
-
- Neuralink (2 shared issues)
-
- 7500 people signed a petition (1 shared issues)
-
- ACX (1 shared issues)
-
- African Gray Parrots (1 shared issues)
-
- Alex Tabarrok (1 shared issues)
-
- Animal Cognition (1 shared issues)
-
- apes (1 shared issues)
-
- Are We Smart Enough (1 shared issues)
-
- Are We Smart Enough to Know How Smart Animals Are (1 shared issues)
-
- Asimov (1 shared issues)
External Links
Source Context
Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.
That's taken up most of my time over the past six months. Going back to blog posts like this is a strange feeling. I wondered if I'd enjoy the break. I didn't particularly; it felt at least as much like trying to resist an addiction as it did resting from a difficult task. There's so much left to say! I never got the chance to tell you whether the SSC Survey found birth order effects to be biologically or socially mediated! And the predictive processing community is starting to really chip away at the question of why psychotherapies work - I need to explain this to someone else before I can be sure I understand it! I only discovered taxometrics a few months ago and I haven't talked your ears off about it yet - that will change! I made predictions about Trump - now that he's come and gone I need to grade them publicly so you can raise or lower your opinion of me as appropriate! And there's the book review contest! We are absolutely going to do the book review contest!
Besides the core insight about the evolved nature of cognition and lots of examples of what that looks like, I also had a few other takeaways about how to do science, the difficulties of sharing discoveries across disciplines, the continuity of neural anatomy, what we can learn from less closely-related brains, and what implications it might have for predictive processing theory.
Besides the delightful stories about animals surprising researchers with how clever they can be (they really are a lot of fun), I found the biggest takeaway to be that cognition is not in a special class, and it shows the same evidence of conservation and continuity that you find everywhere else in evolved beings. Just as all vertebrate skeletons show a clear family resemblance, so too do vertebrate brains and nervous systems. The same underlying efficiency of evolution explains both the continuity and the variation of cognition across vertebrates: if something works, it will be kept around, twisted, changed, and put to new uses (to paraphrase Gibson, "the [environment] finds its own uses for things"). On the other hand, anything that is not helping a species' fitness will find itself on the wrong side of Darwin's balance sheet. This means that you can expect to see great variation in what kinds of cognitive abilities between species, but when you find species with common capacities and common ancestry, they're likely using the same neural lego blocks to do it (and if they're far enough apart, they probably convergently evolved based on similar selective pressures). III. Other Takeaways: How to Do Science, Continuity, Comparison, and Predictive Processing Besides the core insight about the evolved nature of cognition and lots of examples of what that looks like, I also had a few other takeaways about how to do science, the difficulties of sharing discoveries across disciplines, the continuity of neural anatomy, what we can learn from less closely-related brains, and what implications it might have for predictive processing theory.
Birds and cephalopods are especially interesting for understanding cognition and intelligence precisely because they are smart with different hardware. Bird brains still make use of the familiar vertebrate architecture, but the optimization pressures of flight have caused them to have much lighter brains with more neurons per volume. So crows and parrots can be spooky smart despite having brain to body mass ratios that don't look so hot on first blush. Cephalopods are even weirder and harder to understand. They have a lot of their nervous system spread throughout their body, including a separate ganglion for each sucker, so they might have a very different method of cognition. This difference is what makes them simultaneously potentially very valuable for exploring the full scope of cognition in the natural world, but also makes it hard to even come up with good experiments to figure out what's going on inside their rubbery, color-changing non-skulls. Link to Predictive Processing Okay, so the book doesn't actually talk about predictive processing as such, but since it was about cognition, I couldn't help but think about it. Full disclosure, almosteverything I know about Predictive Processing I learned on Slate Star Codex, so take this with a grain of salt. Given that predictive processing purports to describe what is happening at a really low level, if correct, it almost certainly is how other vertebrate cognition works. This might open up new avenues for experiment and observation that aren't open with humans (no, this isn't a call for vivisection or anything, just building on what has been described above about animals having widely varying cognition that can be tested in different ways). Perhaps even more interesting would be to try to figure out if predictive processing explains the cognition of non-vertebrates like octopodes. I don't really have much further insight on this, but this sounds like yet another place where the fields of (human) psychology/neuroscience and animal cognition could usefully inform each other.
It should be impossible for a shared oscillation to have qualia according to one subject, but none according to the other. But honestly, what makes this explanation of qualia persuasive to myself is as subjective as they are. Like my introspection about my information processing makes more sense since I learned about predictive processing, so my introspection about my conscious experiences makes more sense since I understand them this way. For example, while I learn to perform a task it is quite conscious, but once it becomes routine, I can be unsure whether I have just done it. This makes perfect sense now: in accordance with Hebbian learning, the learning process requires repeated synchronous firing, i.e. oscillations, i.e. capacity for conscious phenomena. But the synaptic shortcuts this creates necessarily reduce the bandwidth required for the task, obviating large oscillations with their conscious phenomena.
Inline links: predictive processing, Hebbian learning