Strategic Vision

By Jessica Lake

Our exploration begins with generation, perception, and simulators. It ends with semantic memory, reasoning, utility, communication, and the possible purpose of Strategic Vision.

My goal is not to explain the brain. I have not a clue. But as an EE and CS, I do understand information theory, lossy compression, massive parallelization, and signal flow. I figure if I can walk away with something I could build, I am happy.

Any plausible mechanism is more valuable than reliance on magic.

The Simulator

Let us define a couple of terms I use throughout this discussion.

• Here, reality refers to the external world that you navigate.

• Strategic Vision refers to the integrated internal world that you experience while navigating it. It is the brain’s reconstruction of reality from information supplied by the nervous system.

The Local Generation Insight

A long-held assumption of mine was that a central system predicted reality.

I challenge that assumption here.

My fundamental premise is that the nervous system is highly distributed, massively parallelized, and utterly asynchronous. The latency imposed by a strictly centralized architecture would make us prime candidates for natural deselection.

The nervous system relies upon generation to reduce residual latency. Furthermore, generation allows us to compress reality down to generation error.

Our senses may use local generation loops to anticipate incoming information. The differences between what is generated and what arrives provide the errors needed to improve the next generation.

Locality accelerates processing. Decoupling permits specialization.

The supposed need for unification arises when specialized systems favor incompatible actions. This does not necessarily require a central intelligence. It requires an arbitration mechanism capable of weighting competing responses according to confidence, context, urgency, expected value, and bodily state.

Note: This is a mechanism, not intelligence.

The Problem of Strategic Vision

This led to a major anomaly.

Given that generation is local, necessary, and perhaps sufficient for much immediate behavior, why undertake the bone-breaking labor required to construct Strategic Vision?

A panoramic integration of the senses appears to be major overkill—far in excess of what is necessary for actionable generation.

Action requires:

• Object Recognition

• Predictive Property Sheets

• Projected Motion

• Relationships

• Affordances

Why construct Strategic Vision when decomposition is mandatory for action to occur?

Action cannot operate directly upon the integrated scene. The scene must be broken into objects, properties, relationships, possibilities, and competing responses.

On the face of it, Strategic Vision wastes valuable resources.

The Homunculus Loop

This led me to the Homunculus Paradox.

The decoupled senses generate and sample the outer world. They pass information, including mismatches and changes, onward through the system.

Now suppose Strategic Vision contains actionable information intended for consciousness.

The conscious observer becomes a homunculus: effectively, a tiny person whose nervous system must once again decompose reality and return the resulting information to another observer.

We are trapped in a loop.

If the initial decomposition, local generation, and arbitration are sufficient to navigate much of the world autonomously, then consciousness need not examine a completed inner picture before every action.

This circumvents the Homunculus Loop.

But it leaves me with a problem.

We appear to have a very expensive Strategic Vision that serves no obvious operational purpose.

Aligned Strategic Vision

Who does Strategic Vision serve?

In my overarching theory, consciousness often sees the world in the rearview mirror. It receives much of what has already been processed and becomes the decider of last resort when ordinary mechanisms fail to agree.

Among its more important responsibilities is spoken language.

Perhaps one evolutionary value of Strategic Vision is that it provides corresponding reference spaces between minds.

Humorously, Strategic Vision does serve a homunculus.

It is the person you are talking to.

Strategic Vision is of immeasurable value in face-to-face exchanges. Under these conditions, crude language may suffice.

Suppose you are out walking with your buddy. Suddenly he shouts, “Heads up!”

That is certainly not useful.

But suppose he points as well.

Now we are talking.

You know what he means and which way to jump.

Let us next observe a herd of wildebeests. One senses a predator. Its unease alerts the herd. The others glance in the direction it faces and follow its line of sight.

Suddenly, the herd is alerted and ready.

A face-to-face exchange involving gesture cannot exist in the same way over the telephone.

Try nodding “yes” into your smartphone.

Natural selection, I would submit, favors social groups capable of exchanging a mountain of information with minimal effort.

What you convey is an X on the map.

You study your map. You point. Your buddy locates the corresponding X on his own map.

The maps are not identical, and they are not literally shared. But they are sufficiently aligned for the gesture to work.

As they say, “A picture is worth a thousand words.”

It is remarkable that such a parsimonious gesture can deliver such an outsized result.

Procedurally, face-to-face communication operates within a mutually perceived environment.

1. You recognize an interesting object within your Strategic Vision.

2. You encode and transmit a reference to it.

3. The receiver decodes that reference within their own Strategic Vision.

4. Their attention shifts toward the corresponding object.

5. Sufficient mutual understanding is achieved.

Why So Much Detail?

Think of what is recognizable on your map as a vocabulary.

The right word is terse, precise, and specific. The more we depend upon aligned maps, the larger and more exact that vocabulary can become.

The clarity provided by Strategic Vision, combined with a simple reference, unlocks a vast amount of mutual understanding.

Understood this way, our inner world is not necessarily overkill.

Decoupling Strategic Vision

Changing the role of Strategic Vision really upsets the apple cart.

I have decoupled consciousness from immediacy and from most routine decision-making.

This remains:

1. Generation and correction remain distributed.

2. Much learning remains local.

3. Many action decisions remain local.

4. Conflicting actions require arbitration.

5. Some information becomes broadly available to consciousness.

What purposes might Strategic Vision serve?

• Experiential Awareness

• Communication

• Joint Attention

• Reference

• Episodic Reliving

• Scene-Based Imagination

• Embodiment

The overall architecture has grown simpler.

I have relaxed the stringent constraints placed upon Strategic Vision. It need not be fast enough or precise enough to control every movement directly. It merely needs to be useful to consciousness and to the systems that depend upon conscious access.

Once Strategic Vision was relieved of many of its operational responsibilities, another question emerged.

If Strategic Vision provides an experiential workspace, what role remains for semantic memory?

Simulating Strategic Vision

This raises the question of how Strategic Vision might be repurposed.

Reality is very much like an immersive amusement ride.

As a thought experiment, suppose a Strategic Vision simulator has three levers.

• The Blue Lever controls experiential time. The farther upward the lever moves, the farther into a possible future the simulation extends. Pushing it downward reconstructs a possible past. This permits experiential recall and scene-based planning in minds capable of producing them.

• The Red Lever controls realism—how strongly the simulation adheres to the real world. In everyday perception it is strongly engaged. In a nightmare it may be largely disengaged.

• The Green Lever controls how much volition can direct the action. Deliberate imagination may permit substantial control. A nightmare may permit almost none.

We flagrantly violate physics with our imaginations, yet we recognize the resulting experience as our own.

Here are some possible uses of Strategic Vision:

• Perception corresponds to high realism, present time, and limited volition.

• Scene-based planning favors realism while permitting greater volition and future projection.

• Episodic reliving favors realism, limited volition, and backward projection through time.

• Daydreams permit substantial volition while relaxing realism.

• REM dreams relax both realism and voluntary control.

These capacities need not be equally available to every mind.

A person may reason, plan, communicate, and adapt without voluntarily constructing a vivid internal scene. Strategic Vision is therefore not the universal medium of thought.

It is one cognitive resource among others.

Embodiment

Communication explains why aligned experiential spaces are useful between people.

Embodiment suggests why an integrated experiential space may be useful within a person.

I suggest that we experience higher-order emotions through the integration of multiple systems.

• Terror is not visual.

• Terror is not auditory.

• Terror is not tactile.

Nor is it merely the pounding heart, tightened muscles, altered breathing, temperature, or visceral unease supplied by interoception.

Terror lives at the intersection of all these.

This suggests that full episodic reliving may depend upon reconstructing an integrated experiential state.

One may retain facts about an event without reliving it. The event can survive semantically even when the original sights, sounds, bodily sensations, and emotions cannot be regenerated.

Strategic Vision may therefore distinguish knowing that something happened from experiencing it again.

Semantic Reasoning

A fuller treatment of semantic reasoning appears in my earlier paper, Semantic Reasoning. There I argued that reasoning can be understood as language operating under progressively stricter constraints, eventually taking the form of algebraic and logical structures.

For present purposes, it is sufficient to note that semantic memory stores more than facts.

It stores concepts, relationships, rules, constraints, and reusable structures. These become the primitives consumed by reasoning.

This raises a different question.

If semantic memory supplies meanings, where do those meanings originate?

My earlier answer was that they arise through the compression of Strategic Vision.

I now think that is too narrow.

Strategic Vision may provide an extraordinarily rich source of experiential material. Stable abstractions, relationships, emotional associations, and useful structures can certainly be extracted from an integrated simulation.

But semantic knowledge can also arise through language, instruction, inference, internal deliberation, and the recombination of structures already stored in semantic memory.

Meaning need not begin as a picture.

It need not begin as a relived experience.

A semantic mind can generate new knowledge from relationships among existing concepts. It can reason forward without first constructing a scene.

Strategic Vision provides one source of structure.

It is not the only source.

The problem is therefore no longer simply how reasoning operates, or how meaning is compressed from experience.

The larger problem is how useful structure is distilled from every available source and made durable enough to be reused.

This leaves us with a more differentiated architecture.

Strategic Vision constructs an experiential world.

Semantic cognition extracts, preserves, and generates durable structure.

Other mechanisms use that structure to select responses, resolve constraints, and guide behavior.

None of these is the homunculus.

None is the whole mind.

The important observation is not that the brain works exactly this way.

It is that it plausibly could.

No magic is required.