Anyone's
Intelligence

In my hypothesis, foundation refers to innate mechanisms with genetic and physiological bases. Hunger drives feeding, and tendencies to seek benefit and avoid harm influence responses. These provide the basis on which acquired experience acts; not every drive must first be explained through another memory.

In the sweets example, avoiding pain has an innate basis, whereas the association between sweets and something to avoid involves acquired experience. I hope to draw on the joint action of these foundational mechanisms, learning, and memory to implement simulated cognition in a model.

When I understand something, I need to draw on other memories and find connections that can explain it. I use Cofoundational knowledge to express this process: new information is interpreted through existing connections, and existing memories contribute to a new understanding.

Feelings and language continually influence one another. A word connects a past experience to what is happening now, changing the current feeling; that feeling then influences which contents are more likely to contribute to understanding. This ongoing interaction is a cognitive process I hope to simulate in a model.

Imagine someone being mocked for being fat. In my hypothesis, physiological changes involving cortisol, adrenaline, and related processes contribute to memory encoding. The experience acquires a higher base weight and becomes easier for relevant cues to evoke later. This is one step in connecting experience, the body, and memory.

I use “base weight” to describe the memory's tendency to be recalled later, and “base color” to describe the feelings it carries. Here, that base color may be anxiety or fear, associated with responses that avoid pain.

If someone already remembers being mocked for being fat, then learns that “sweets make you fat,” the word “fat” becomes a connecting point. The term “word” also points toward tokens in a model. When they encounter sweets later, the relevant information may follow this connection and evoke the memory of being mocked.

Imagination / analogy This process reminds me of a Transformer: through their relations to other content, words / tokens influence how much relevant memories contribute to current understanding. Connections between memories and their weighted contribution are what lead me to this analogy; specific correspondences with model components remain to be explained.

A memory has a base weight formed through experience, while the association between current information and that memory further influences how much it contributes. I use “multiplier” to describe how the relation acts on the base weight. In this example, the remembered causal association between “fat” and “being mocked” makes the relevant memory more likely to contribute to the current process. Whether this remembered causal association matches external facts is a separate question.

I understand activation as a matter of degree. When someone faces a sweet, experiences related to eating and memories of being mocked can contribute together. Weights influence the degree to which each connection contributes.

Imagination / analogy This also reminds me of a Transformer, from a different angle than the preceding section. That section focuses on connections between words / tokens and relevant memories; here, the focus is on the weighted, joint contribution of multiple connections. How the current state modulates this contribution is a relationship I hope to simulate in a model; this account does not yet map it to specific Transformer computations.

The current physiological state also modulates this process. When someone is very hungry, the drive to eat is stronger; when the memory of being mocked is strong, avoidance may exert more influence. Behavioral choice is therefore state-dependent: the same person facing the same sweet can reach different outcomes. I hope to use this state dependence as a reference for quantifying and simulating the joint influence of memory weights and current state on a model's choices.

I use the x-axis to represent how a chain unfolds, without specifying a unit yet. I also use it to represent a person's history of repeatedly going through these processes. The y-axis represents abstraction: lower positions are close to foundation, while higher positions are more cofoundational, involving more intermediate words and connections and longer, more complex chains.

Imagination / analogy In a two-dimensional plot, I imagine a curve starting low, rising through abstract relations, and returning to physiological responses and behavior at a lower level. The body and abstract understanding can then be traced along the same chain. The curve's rise and fall express this idea; they are not a function that has already been derived.

I borrow the names Symbolic, Imaginary, and Real to divide the y-axis into three broad levels. The Symbolic is highest and most abstract. In this example, “sugar—mockery” expresses an abstract relation, while “fat” connects contents from different experiences.

The Imaginary connects the upper and lower levels. “If I eat sweets again, I will be mocked again” places that relation in a situation involving oneself. This imagined situation can produce fear and physiological changes, influencing whether the person continues to eat sweets. The Real, as I use it here, means the concrete bodily states and responses of the present. The chain therefore passes through words and relations, through imagination involving oneself, and back into physiology and behavior.

Different memories connect through shared words or relations, represented in the plot as intersecting curves. “Fat” connects eating sweets with being mocked, allowing contents formed in different situations to contribute to the same response.

These connections can occur at different levels and gradually form a network. Weights and the current state jointly influence how much each connection contributes. Intersections represent these connections, while some intermediate processes are omitted for now.

Foundation and the Real are close, but they are not the same. Foundation consists of innate mechanisms; the Real is a person's concrete bodily state and responses in the present. That person already has experiences and memories and is influenced by imagination, so their current state carries a history.

Imagination / analogy I have used “the infant” and “a person after ‘castration’” to express this distinction. In my diagram, the latter more directly means someone whose history already includes many such chains. These are metaphors that help me articulate the idea, not grounds for deriving additional developmental stages or conclusions about Lacanian theory.

Whatever abstract relations a chain has passed through, it ultimately returns to the body and behavior. It does not thereby return to a state untouched by experience. This is why I place the Real near foundation without merging the two.

This network looks very much like a Transformer to me. A word can engage other contents, different connections carry different weights, and those weights influence what contributes to the present response. For example, “fat” makes a memory of being mocked easier to recall. Examples like this are how I began connecting my imagined memory network with language models.

I want to preserve this sense of resemblance and keep asking: what correspondences and differences might exist between the words, memories, connections, and weights involved in understanding and those in existing language models?

“A rational virtual machine running on an emotional host operating system” and the movement from “the infant” to “someone whose history already includes many such chains” helped me articulate my questions. “The linguistic structure of the unconscious” then suggested a route connecting them. I want to preserve these ideas as they emerged while leaving room for further explanation and revision.

Putting the preceding ideas back into one example: someone has innate mechanisms for feeding and for seeking benefit while avoiding harm, and has also been mocked for being fat. In my hypothesis, physiological changes at the time increased the memory's base weight and left anxiety and fear as its base color. Later, they learn that “sweets make you fat,” and “fat” connects this statement with the earlier experience.

When they encounter sweets again, the “sugar—fat—mockery” connection contributes to the response, with a relation multiplier describing how these connections act on the memory's base weight. They may imagine, “If I eat sweets again, I will be mocked again,” producing fear and associated physiological changes. The innate tendency to seek benefit and avoid harm acts on behavior through acquired connections. In the example I am imagining, these effects ultimately lead them not to eat the sweet.

Other connections continue to contribute. If the person is very hungry, the drive to eat also increases. Having this memory therefore does not mean they will always refuse sweets. I hope to draw on this process to quantify each connection's contribution and its variation with the current state in a model, as part of simulating cognition.

This account has connected innate mechanisms, experience, memory, language, imagination, and bodily responses. Further explanation needs to clarify how different weights and physiological states relate, and how the axes represent abstraction and successive rounds of experience. The items listed here have not been fully developed on this page; that does not establish whether I have already formed ideas about them.

Discussion has also raised questions about initial memories, whether imagination feeds back into encoding, and the distinction between understanding and misunderstanding. These remain questions awaiting my response, not premises I have necessarily accepted. No clearly established self-contradiction has emerged in the discussion so far, but the consistency of the whole hypothesis and its empirical basis still need to be investigated separately.