What's an egocentric reference frame supposed to be in abstract space?

It is clear that an egocentric reference frame in the physical space is attached to a body part or patch of a body part.
But for the case of one for abstract concepts such as language, not so much.

Egocentric reference frames are responsible for translating between different allocentric locations to enable communication between the different columns that represent possibly the same object.
This makes egocentric reference frames seem like a fundamental concept to me.

I was thinking about it and I wasn’t able to come up with an interpretation that neatly explains everything.
My best guess is that it has something to do with attention; the abstract egocentric reference frame is located at where the agent is attending to in the abstract space.

Any idea from the forum? :thinking:

1 Like

Consider the postural system as the mapping between ego and allo centric physical spaces.

When the agent has to reach for/look at something, the system has to transform the allocentric location into the right egocentric location.
Is this what you’re trying to say?
I don’t see how it relates to what it means to have an egocentric reference frame in an abstract sapce…

Egocentric implies that there is a relevant concept of the self in a given space. In language, that might correspond to the I or me concept. Or if you are reading a story, it might be the point of view of whatever character or narrator is telling the story.

Egocentric and allocentric reference frames only make sense if a discernible relationship exists between the self and the other. Perhaps ego- and allocentric are not the proper terms to use when referring to abstract spaces.

Any continuous space should have a concept of origin and/or unit. The unit is the yard stick by which you measure the distance (or similarity/difference) between two entities. The origin is the place with respect to which you are measuring. So, in physical spaces, we use egocentric coordinates when we are the origin (when we are measuring distances with respect to our body), and allocentric coordinates when we are measuring distances with respect to some other entity. The origin can be shifted, but the unit must be relatively fixed in order for comparisons to be useful; otherwise some form of unit conversion (comparisons of units) must be utilized.

In abstract or discrete spaces in which there is no reliable sense of continuity, direction or distance, then the reference frame concept gets a little harder to define. If the space can be visualized as a network of nodes and edges, then the origin can be associated with any node, and distance can be associated with edge lengths or weights integrated over the allowed paths from one node to another.

TLDR: Egocentric and allocentric coordinate frames are special cases of general reference frames. General reference frames will almost always have a concept of origin and unit. The origin and unit are typically well defined in continuous spaces, but the concepts can also be extended to more abstract spaces.


My takeaway from that is that voting for possible objects through egocentric reference frames is not general and doesn’t always extend to abstract cases.
Is this interpretation correct?

Yes. That’s more-or-less the gist. Just realize that the egocentric origin (self-centered) is just one of many possible origins, and that origin is only well defined in a very limited number of circumstances. The appropriate reference frame(s) is (are) likely going to be established by and anchored to the most salient features and/or most relevant context(s).


For a very simple one layer system this is generally correct.

In the brain the WHAT/WHERE streams become more abstract as you ascend the hierarchy. By the time you get to the areas around the intersection of the parietal and temporal lobes you see processing like mental rotation of physical objects.
The abstraction you are looking for are properties of a more complex system than is generally state-of-the-art in current HTM implementations. The theoretical foundations to do these things are not in place yet but as soon as people start exploring the H of HTM I expect progress on this front.

1 Like

Egocentric reference frames can be far removed from body parts and the sensory surface, I think. Hemineglect shows an example. It’s neglect of the left or right side of the world, not always the exact same as the halves handled by the brain’s hemispheres. I imagine egocentric spaces are learned.

Egocentric spaces can help tie down the dimensions used in allocentric regions and abstract regions (I’m speculating here).

In allocentric regions, there are many object spaces but I think use the same dimensions in each region, e.g. 2d or 3d. That helps link objects into bigger one and reuse learned things (like how to move through allocentric space).

Different dimensions will be more useful for different objects, making it harder to pick just one set of dimensions during learning. Egocentric regions don’t have that problem, so they can help.

Likewise, egocentric regions can help tie down dimensions in abstract regions. If a set of dimensions is a way of thinking about a concept, it’s important to use a limited number. That helps link concepts together and reuse knowledge.

Things like language aren’t in abstract space themselves, but they can mirror abstract space with sensory/motor representation (e.g. hearing your own thoughts). Whereas concepts can be thought of in a bazillion ways, the equivalent representation in more tangible forms helps tie down the ways of thinking, because relatively consistent ways of thinking can be derived from those tangible representations.

Language is a way to represent abstract things in a unified way. It converts concepts to a sensory/motor representation. Hearing thoughts in your head feels like speaking and hearing. I imagine it’s easier to keep track of and grasp your own thoughts if you form a sensory representation.

There are other ways besides internal monologue (which some people don’t have whatsoever). Gut feelings seem like a way of representing abstract thought using subcortical emotions. It can be a literal gut feeling because people can feel different emotions mapped onto different parts of the body. Another way of thinking is in terms of actions.

While I mentioned the importance of distribution of abstract meaning through the hierarchy, I neglected to point out that this distribution is grounded at the egocentric level.

This paper shows the parts of the brain most closely associated with the personal perception of semantic concepts and how those egocentric portions of the brain grounds the semantic meaning. I fully expect that this grounding is extended through the hierarchy as experience with the concept is acquired.

The " Similar articles" section of this page has some very interesting looking links. So many papers - so little time.


One way of thinking about this in a high dimensional space would be to see an egocentric view as immediate navigation eg from a specific concept there are a limited number of direct associations that can be made. Navigating the space is going from one state to the next.

The allocentric view would then closer to having a map. This would allow switching between perspectives that are not immediately related.

Allocentric might be associated with the slow rational abstract thinking while egocentric allows for rapid sequences e.g. like following a particular route when driving to a familiar destination.

1 Like

@markNZed Are you saying something along the line of this?

If so, what would you think the implications be, other than things you’ve said so far?

No, attention is what the agent is attending to, the perspective is egocentric or allocentric i.e. when I imagine I’m walking across the room I’m attending to the imagined trajectory from an allocentric perspective. When I am walking across the room I’m attending to the objects along the trajectory from an egocentric perspective.

I just published a book called The Esoteric Helen Keller. I embedded a cognitive philosophy in the middle of that book which discussed a hypothesis I call Navigational Consciousness. The discussion of an Egocentric versus an Allocentric mind is central to the hypothesis. I recently shared a summary of the book with a neuroscientist and will attach it to this email. Food for thought from outside the engineering world.

(Attachment Excerpts from the Esoteric HK.docx is missing)

If anyone here hasn’t, I’d recommend Wayfinding by M.R. O’Connor.