The Tabletop Illusion

I'm currently reading the soul fallacy by Julien Musolino. In it he tries to stress that we are constantly being fooled by our subjective first person perspective and in our judgements. That we need the objective third person perspective as provided by science to tell us how reality really is.

He tries to illustrate his point by the following perceptual illusion of two tables (the Roger Shepard tabletop illusion).

I think we would all agree that the two tables are different shapes. But, when we rotate the left table 90 degrees clockwise and move it over the other table, we find they are both exactly the same shape! Moreover, even when we are made aware of this fact we cannot help but see that the tables are very different shapes. Julien Musolino finds this very significant. In the the soul fallacy he says:

[This illusion] pits our first-person perspective—what our senses subjectively reveal to us—against the third-person perspective—what the result of objective measurement demonstrates. The story of the demise of the soul, to a large extent, reflects the triumph of the third-person perspective over its subjective, first-person counterpart. But there is more to Shepard's illusion than meets the eye. The analogy contains two additional virtues. After this brief demonstration, I do not know anyone who would continue to insist that one tabletop really is longer than the other because of the way they look ... The analogy's second virtue is that it reveals to us how stubborn first-person impressions can be in the face of objective evidence.

I think this is flawed reasoning. Indeed, it seems to me to be inappropriate to call this an illusion at all.

First of all, we need to draw a distinction between the image of the tables, and the actual tables themselves that they represent. The above picture is an image of two tables.

Having put that aside, let's suppose there are two tables in front of us that are in our visual field. When looking at the tables, let's also suppose that the image of the tables on our retinas closely approximates to the image of the tables we see above on our computer screens. Just as the shapes of the images of the two tables is the same on our computer screens above, so too will the shapes of the two images of the tables on our retinas be the same.

Are we fooled in the latter case? I would say not since if we approached the two tables, look at them from different angles, run our fingers over them and so on, we would definitively establish the two tables are of differing shapes. Likewise, if we are to take the drawing of the two tables above as actually depicting tables rather than arbitrary lines representing nothing, it is not fair to say we are being fooled here either. In other words, there is no illusion, as such.


We need to understand how are senses work. We don't simply passively perceive what's out there. The data we receive through our senses is hopelessly inadequate for us to see the world as it truly is in and of itself. The actual ability to see is supplied by our implicit expectations gained from our previous acquaintance with the world. That is to say, what we actually see is shaped and moulded by all of our previous visual experiences. In effect, we have an implicit unconscious theory about how the world is and this is instrumental in shaping what we actually perceive.

Without such an implicit theory, we wouldn't be able to perceive at all, at least not a 3D world of objects at various distances. We'd just see a splodge of colours in a 2-dimensional plane. In other words, we'd see the world as a computer or robot would; namely as depicted from a third person perspective shorn of any "illusions" that any minds can add.

In short, this "illusion" is not an example of us being fooled. If we didn't experience such "illusions", then we wouldn't be able to see at all! This is why autonomous cars -- which do not have the benefit of illusions to apprehend the environment correctly -- shouldn't rely upon cameras alone. They need other instruments to effectively detect the environment, such as LIDAR. Even then, they are not as proficient as human drivers in urban environments. My personal expectations are that fully (level 5) autonomous vehicles will not be widespread for a few decades yet (back in 2014 the date I gave was 2060).


So, is the subjective first person perspective inferior to the objective third person perspective as Julien Musolino claims? Certainly not when it comes to visually apprehending the environment. So much for, as he puts it, "the triumph of the third-person perspective over its subjective, first-person counterpart". This is not to deny that we are often fooled and that the third person perspective, as provided by science, is the more accurate. But I don't think this has any relevance for deciding whether, on the one hand, our essence is a soul or, on the other hand, a sophisticated biological machine. But I'll address that issue when I come to reviewing his book.

Two other relevant similar blog posts by me are:

Are Perceptual illusions always necessarily illusions?  (essentially the same argument I make in this blog post, but I wrote it over 9 years ago).

Perceptual illusions show our minds construct reality