Making Sense of the Cognitive Shuffle
There are several big ideas behind mySleepButton’s current and upcoming cognitive shuffle packs (as well as DIY versions). I’ve written some papers about these ideas before, and am currently co-authoring a new paper with Célyne H. Bastien on the subject. The theory we are developing is called the “somnolent information-processing theory“.
This theory proposes that the brain hosts a virtual machine that we call the sleep onset control-system. This virtual machine controls the entry into sleep-onset (SO). Contrary to popular opinion (and to a received misnomer in sleep research), normal SO does not involve a “sleep-state switch”. The reason it does not contain such a switch is a rather nice illustration of cognitive science, as it is an interdisciplinary argument involving AI, philosophy, psychology and neuroscience. The empirical evidence for this comes from papers authored independently by Tadao Hori and Robert D. Ogilvie. They (independently) argue that SO is a process. For Hori, it is a nine stage process (not a binary switch!). We will review that some other day.
For now I want to focus on another big idea behind the cognitive shuffle. It is that the SO process involves a meta-cognitive shift during which the virtual machinery that detects incoherence in working memory (and related transient states) weakens. The brain becomes even less concerned with maintaining mental coherence than it normally is.
One of the interesting implications of somnolent information-processing theory is that to understand SO one needs to understand the functional architecture of the human mind, and diverse mental mechanisms. Thus, although SO is a very brief period, to understand it requires a broad and deep understanding of the mind. This partly explains the lack of progress in insomnia research, and why insomnia remains such a pervasive societal problem. It is not merely a problem of “knowledge translation”. It is a problem of lack of scientific knowledge to translate in the first place. This, in turn, is due to the (largely unrecognized) difficulty of the problem of understanding insomnia.
If Célyne H. Bastien and I are right about causally relevant meta-cognitive features of SO, then to understand SO requires an understanding of how the brain detects and resolves incoherence. This calls for an understanding of coherence/incoherence. There are tough problems in cognitive science that require a grounding in AI.
Yesterday, on the CogZest blog, I posted the following article:
This is an overview of a presentation I will give at an upcoming local humanist meeting.
We, humans, are designed to try to make sense of our experience. Coherence is deemed to be necessary for rationality. Rationality is a fundamental principle of humanism. However, it is impossible to ensure that the various models of the world, which we construct and carry with us, are coherent with each other and the world.
The “We(e)” in the title, of course, is a double-entendre, we are wee sense-makers. We try to make sense of ourselves and the world, but though some of us get further than others, none of us gets very far. Or as Popper put it “While differing widely in the little bits we know, or rather guess, in our infinite ignorance we are all equal.” I might add “in our incoherence we are all equal”. (However, we don’t have metrics of incoherence. That would call for better models of the mind.)
AI teaches us something that is difficult if not impossible to understand without it: Not only is it difficult to know what parts of our knowledge might be false (as is commonly recognized), it is difficult to detect inconsistencies in our own knowledge. If you’ve ever double-booked yourself for lunch, or looked for something that was in your pocket or even in your hands, then you’ve had an experience that illustrates this fact.
It is not far-fetched to claim that at SO our concern for coherence wanes. But to explain this in detail is more difficult.
The CogZest blog post quoted mentioned above provides a few pointers to concepts and literature that are relevant to understanding coherence. But they are just a few.
For an easy to read introduction to cognitive science, including functional architectures, see Cognitive Science: An Introduction to the Science of the Mind by José Luis Bermúdez of Texas A & M University.