What follows is, as the title says, a revisit, revision, and expansion of an earlier post I made. I may continue doing this as new thoughts come to mind while I work through my thinking on the subject of Philosophy of Mind.
Understanding how consciousness and the mind is generated is best done using the bottom-up approach of neuroscience. But if the consciousness/mind is performing recursive, downwardly causal actions on the intentional object – the content of thinking/cognition – then what are the mental mechanisms being utilized? Here I present some nascent ideas for your consideration.
Categories of Understanding and Laws of Thought. These are the template for which – or the form/structure through which – understanding occurs. This might be analogous to converting a keystroke on your keyboard into a particular computer language. I wrote about my theory in this post. This is the Kantian idea that reality itself is created in our brains using empirical data taken in through the senses; the categories are the building blocks of that reality. In the post I linked to, I split the various categories up into three general categories: primary, descriptive, and prescriptive. Primary has to do with the self and its relation to the world (I vs not-I, mine vs not-mine); descriptive has to do with the more passive ordering of sense information (big vs small, distant vs close, etc.); prescriptive is the active attribution of elements to sense information (agency vs inanimate, valuable vs worthless, related vs not-related, etc.).
Different people employ category templates in different ways, which causes different understanding of the same concept between individuals. The first act of intuition, upon perception is to fit the perception into the Categories of Understanding – basically, to construct the ‘shape’ of the intentional object using the categorical template.
Qualia. This is the palette used to build the content of intentional objects into the structure/form of the categories of understanding. The categories determine the shape of the content of thought, but qualia is the material from which the content is built. Or, in staying with a computer analogy, the categories of understanding are the particular computer language being used and qualia is what the data being shuffled around using that language.
Values. These are the way one relates the self to the intentional objects that are generated by the composition of Categories and Qualia – of form and substance, so to speak. This is a necessary step, because the intentional objects independent of a relationship to the self are non-existent entities. The value-relationship of intentional objects to the self, however, alters the intentional idea from any sort of pure form.
There isn’t as good of a computer analogy for values as there is for categories of understanding and qualia, since computers don’t value things – a computer might have compatibility issues with a certain computer language, but it doesn’t care; none of the data stored on a computer is viewed as more important than any other data.
The mind, also, seems to have a way of calling forth and pushing away intentional objects into conscious consideration. It’s then on this conscious stage, with a size determined by what’s called working memory, that the mind can consciously (or unconsciously) perform various operations on the intentional objects.
But what are the mental ‘instruments’ being employed to manipulate the intentional objects inside the mind?
I suggest what I’m going to call the Operations of Intuition: the things the mind does to its internal model(s) of an intentional object when it is actively and consciously undergoing thinking/cognition about that/those intentional object(s). Being able to understand these Operations of Intuition may be vital to the development of an artificial general intelligence (AGI).
I want to be clear that these Operations of Intuition are not homunculi but perhaps better understood, in the computer analogy, as programs composed of various subroutines. In this way, when the brain employs them on intentional objects, a particular pattern of neuronal activity is engaged. Below I will explain the Operations of Intuition in a phenomenological sense rather than a neurophysiological sense.
Here are my ten candidates for what the Operations of Intuition are:
1) Set goals. Goals are a state of affairs we want to make a reality through the manipulations and employment of our intentional objects. It’s the attribution of functionality to the intentional objects – how can this intentional object most efficiently realize the state of affairs I desire? A state of affairs may be something external – get the job I want – make dinner, scratch an itch – or something internal – wrap my head around a difficult concept, make myself happy, or become aware of the time.
2) Pattern Recognition: determining patterns (internal similarities) within intentional object(s). This is essentially the function that neural networks perform with given data sets.
3) Abstraction and Symbolization.
a) Abstraction: apply and adjust the species and genus of the object’s classification by comparing to similar known or conceived intentional object in memory or immediately present (external similarities).
b) Symbolization: imbue certain intentional object with meaning beyond what their form and qualia immediately imply. This could be anything from imbuing a cross with the idea of God’s Grace and Christ’s Salvation down to having the squiggly lines (or mouth sounds) that make up the word Pokemon contain the meaning of all the various creatures invented to inhabit this fantasy world.
4) Estimate Various Probabilities.
a) probabilities for things like cause, purpose, property, stability, etc.
b) probabilities for how likely a certain state of affairs is given one’s internal mental model of the world.
c) probabilities for predicting what state of affairs will follow from the current state of affairs.
5) Evaluate Salience and Value.
a) Salience to one’s various goals and to potential new goals. A goal can be anything as simple as getting across the room or reaching for one’s cup of coffee – it’s an action with an intentional outcome.
b) Value to one’s self. This can be seen in a sort of Bayesian learning scheme, where the prior is one’s already held values and the posterior is how new information alters one’s values, such that the output is a set of values altered by a degree characterized by the relative strength of the prior (strength of one’s convictions) (the P(H) in Bayes’s theorem) and the evidence (strength of the incoming data in the opposite direction of previously held convictions) (the P(E|H)/P(E) in Bayes’s theorem).
6) Attribute value judgments (good, bad, safe, dangerous, pleasurable, painful, etc.) based on one’s own value system. This is a triangular relationship of self, objects of various value to the self, and the abstract notions of value as applied by the self to those objects of value.
7) Assimilate, accommodate, or reject novel data to one’s internal models of the world (schema in the Piaget parlance).
8) Separate, combine, mix, extract and/or alter the degree or quality of properties of intentional object(s) being considered. For example, taking the body of a human and combining it with the body of a horse to make a centaur; perhaps then making it pink with purple polka-dots and 50 feet tall; and then instilling it with the abstract Idea of freedom due to its speed and strength.
9) Rotate and translate model(s) of the physical object(s) being considered in space (relative to other objects and to one’s self).
10) Theory of mind – determine the agency of and then apply potential motivations and thoughts to other agents. See also Intentional Stance.
These Operations of Intuition can be employed either subconsciously or consciously. How they are employed is a matter of training. In the untrained mind, attempting to utilize these operations sequentially in order to solve a problem (not necessarily an intellectual problem such as math, but the general problem of “what next?” when considering what to do with any line of thinking – more on that in a bit) they will be used subconsciously in a chaotic fashion. By using language (talking to ourselves or writing it out) we can bring each of these operations into conscious thought as we utilize them. It’s there that we can learn in which order and in which combinations to utilize these operations in order to solve a particular problem (whether we are solving it correctly or incorrectly, so long as our particular way of doing it is reinforced). Practicing the problem over and over will reinforce our way of utilizing the operations (whether correctly or incorrectly), eventually allowing the order to be implemented in an ‘algorithmic’ fashion so as to utilize these operations subconsciously, but orderly rather than chaotically.
Once trained for a particular type of problem, the brain can subconsciously employ the ‘algorithm’ or solve the problem. Multiple problems can be clustered into these ‘algorithms’ of Operations of Intuition. The brain can then call forth multiple ‘algorithms’ for even more abstract and complicated problems. Becoming good at these higher levels is where expertise arises. For instance, a medical intern might come into a room where a patient is crashing and have to consider each step of what to do to stabilize the patient consciously (and therefore slowly) even if they know each individual step well enough to the point of being an ‘algorithm’ whereas a veteran attending doctor would have grouped each of those ‘algorithms’ into a larger cluster and be able to utilize them without much conscious thought.
So, what might be considered a ‘problem’ with which our brains would want to utilize these Operations of Intuition? As I said above, a problem isn’t necessarily some math problem. A problem might be as simple as figuring out how to open a door or pick up the TV remote and turn it on. These things don’t seem particularly like problems because most adults have mastered them enough that the algorithm is done completely subconsciously. But we could see how a few of these operations would be used for something like picking up the TV remote to turn on the TV.
First, I have to decide I want to watch TV, so operations (1) and (6) are used; second I use (4) and (5) in order to know how I go about doing that, of which there are two ways (the remote or the button on the TV), so I use (6) to decide that using the remote is preferable to getting up and using the button on the TV; I use (2), (4), and (5) to decide that the remote will actually work when I press buttons on it; and so I use (9) to know what actions to use and (1) to decide to actually reach over and pick up the remote; when I press a button and turn the TV on, (7) will be used to interpret all of the previous actions.