STM and LTM

Short-Term Memory and Long-Term Memory in the Attractor Framework.

Short-term memory (STM) and long-term memory (LTM) fit naturally into the attractor-based explanation of the Hopfield model and neural dynamics.

STM

Short-term memory is the brain’s ability to hold information for a few seconds or minutes.
Short-term memory arises from temporary attractor states of energy flows and distribution.
In attractor terms, STM corresponds to shallow, temporary attractor states.

Here is how STM works in this framework:

1. A visual pattern activates a group of neurons.
2. These neurons keep each other active for a short time.
3. This creates a shallow attractor basin —stable but fragile.
4. The pattern can be held briefly without being permanently stored.
5. New input or distraction can easily disrupt it.

STM is like balancing a ball in a shallow bowl: stable for a moment, but easily knocked out.

Examples:
- Holding a phone number for a few seconds.
- Remembering a shape long enough to copy it.
- Keeping track of a moving object.

LTM

Long-term memory lasts for days, years, or a lifetime.
Long-Term Memory consists of deep, stable attractor basins.
In attractor terms, LTM corresponds to deep, strongly reinforced attractor basins created by physical changes in synapses.

Here is how LTM works:
1. Repetition or emotional importance strengthens synaptic connections.
2. This deepens the attractor basin in the energy landscape.
3. The memory becomes stable even without ongoing neural firing.
4. The system naturally falls into this attractor when similar input appears.
5. LTM is therefore a structural change, not just temporary activity.

LTM is like carving a deep valley: once formed, the system rolls into it easily.

STM and LTM in Visual Recognition

STM and LTM work together in visual recognition. When you observe lines, shapes, or gestalts:
- STM holds the pattern briefly as a temporary attractor.
- LTM stores the pattern as a deep attractor that can be recalled later.

Recognition happens when STM activity “locks onto” an LTM attractor.
This explains why you can recognize familiar shapes instantly, even when they are blurry or incomplete.

Limit Cycles and Memory

Limit-cycle attractors also contribute:

- In STM, they support rehearsal loops (mentally repeating something).
- In visual processing, they support scanning patterns.
- In LTM, they can encode sequences, such as movement patterns, for example: dance steps or handwriting motions.

Thus, both point attractors and limit cycles help the brain stabilize and recall information.

Image B: Animation of a Point Attractor.

Summary

Certain lines, shapes, and movement patterns are easier for the brain to store because they form stable, low-energy attractors in neuronal circuits of the occipital lobe.
These attractors help the brain recognize faces, objects, and gestures.
They also play a role in biological attraction (during romantic periods), because certain facial patterns stand out more and are perceived as more beautiful.

We find some of these facial patterns attractive, and this helps with pair formation between men and women. This romantic bond can lead to living together or marriage, and possibly having children.

When people with these attractive features reproduce, both the memory patterns (MAP) and the facial patterns (FAP) can be passed on genetically (GAP).
Over many generations, this process can shape both the structure of the face and the way the brain processes visual information.

In this way, attractors operate both inside the brain (as neural dynamics) and between people (as biological attraction). Together, they help explain how perception, memory, and evolution interact.