Perceptual processes: Visual perception

The most important part of our brain in the process of visual perception is the occipital lobe (right in the back of human skull, opposite to our eyes).

The brain is built from tinny little neurons.

We are born with roughly 100 billion neurons. A neuron is a tinny little cell that has a body, a long "arm like" part named axon, and small finger-like dendrites. The axon later in the developmental process develops a whitish color coat (just as the electric wires have a plastic insulating wrapping coat). The dendrites branch our like in fingers and receive signals from other cells/neurons. The information passes from one neuron to the other with help from neurotransmitters (chemicals in our brain). The place where the dendrites contact and pass the impulses using the neurotransmitters are called synapses.
In the developmental process billion of new synapses are formed. Each neuron may be connected to as many as 15,000 other neurons, and form a very complex network (sometimes called "brain wiring").

Some believe that we see illusions in a certain way because of the eye's physical treats, others believe that is due to the culture we develop in.

In general scientists believe that brain and eye work together and it all depends on the perspectives we look at what we see.

The eye sees but it is the brain who will decide “what” and “how” the eye sees.

It seems that we see illusions in different ways, according to how our brains work.



The Gestalt psychologists believed that a number of innate tendencies (Gestalt laws) influence the way we see. Contemporary psychologist agree on the universal laws, and that we perfect and they are influenced by experience and learning.

Proximity - things closer to each other seem to belong together

Closure - we have the tendency to see the entire thing not only a specific part that we see

Similarity - you know the old saying "birds of the same feather flock together"

Good continuation - our brain likes to take short cuts, and the easiest and most familiar path.

When we process information we use couple of most common theories of perceptual processing:

Template matching theory sustains that we have specific set of templates and in the process of perception we try to match what we see with the templates stored in our memory.

There should be a exact, or very precise match between the template and the stimulus. Useful for standardized processes (credit cards, bank accounts, etc.), but is not able to explain the rich and complex stimuli from our daily environment.

Colors and similarities create confusion and problems in perceptual processing.

Feature analysis theory proposes that we compare the visual perceived stimulus with small characteristics called features. However the problem with this is that there is no explanation for relationship between the features. Another problem is that not everything in the human environment is simple and can be reduced to simple features. More complex stimuli need complex perceptual processes. Support from neuroscience and psychological research. Does not explain relationship between complex stimuli. Applies mostly to letter recognition, two dimensional writing.

Structural theory or Recognition-by-components gives a tridimensional (3D) explanation to perceptual processes using the simplest 3D shapes called geons.
It seems that we store specific 3D views so in the case of modified images or modified view point we are less able to recognize objects. Geons can be combined in meaningful objects. In general arrangements of 3 geons seem to be enough for people to recognize an object.

Recognition sometimes requires mental rotation when the object is viewed from a unusual angle.

In conclusion: humans use in the process of visual perception different theories at different times for different purposes. We might use Template matching and feature analysis when we read, and recognition-by-components when we perceive objects and 3D stimuli.
The theories do not negate each other but complement each other and give a more complex (even if not complete) explanation of perceptual processes.

Let's add to this the idea that we do two kind of processes Bottom-up (when we start from the actual visual stimulus and try to process it directly) and Top-down (when we try first to make sense of the perceived stimulus by using what we already know about the item we see, and we base our information processing on previous experience and stored memory).

That is why we can make good sense of ambiguous things we see....
Perhaps that is why we can see an entire story into a blot of ink.




Take a second and look at the picture below (you might know already the drawing); what do you see?


Did you see (when you first ever looked at this drawing) , a young woman or an old woman?
Do you see them both now?
This happens always in my class of approximately 20 people... it is not uncommon that some people will say they see ONLY the young woman, others will see ONLY the old woman; and some others will see both.

The drawing is the same for everyone. So, then why some cannot see both images, and they seem to prefer to see only one type of representation?

This is because of some processes called bottom-up and top-down processing.

Bottom-up processing takes place when all 20 people look at the drawing. They all see it, they see the white and black patterns mixing.

Top-down processing takes place when people try to make sense of the image they see. Based on their predisposition, preference, and mostly because of the previous information stored in their long term memory, some people will see one or the other image. That "making sense of it based on previous knowledge" is in fact the top-down processing.

Bottom-up processing is data driven, starts from the data, stimuli we perceive and is processed thru the physiological organs and by the brain.
In real life we usually start with bottom-up since we process what we see, however almost simultaneous we try to make sense of the stimuli we perceive, and in the process of giving a meaning we do top-down processing (conceptually driven processing). That is higher level mental processes influence the basic bottom-up perceptual processes. We have expectations, stereotypes, and mental sets that bias and form our perceptions (that is why we can tell a story when presented with an ink blot). Our expectations guide our perceptions. Think when your printer runs out of ink and printed only the bottom half of the lines in the document you wanted to print from the web. However you were able to read it. That was because you did most likely lot of top-down processing. You recognize the English alphabet and are familiar with words, so you are able to read even if the letters are blurred or not completely printed. If you were to read a foreign text then the task would be more difficult.

Top-down processing is good and bad! Is good because it saves us lot of energy in the information processing, but is bad because at times it causes us to make mistakes since we become too confident in what we expect to see and we do not pay attention to what is really there to be seen (over active top-down processing). Top-down processes are influenced by our biases and stereotypes. That is why an overactive top-down processing will bring errors in word and object recognition.


According to Parallel Distributed Processing (PDP) theory - connectionism- cognitive processes are understood as networks that link together separate units from processing of stimuli. One item will activate other items that might be linked to the activated one. The network model can be imagined like a fish net with many nods and links between many of the existent nods. If you pick up the net from one little nod part of the net will come with it. That is the same as when a stimuli (let’s say one word) is activated in your memory.

If I write “Jack Russell Terrier” in you mind - depends on how much knowledge you have about what I just told you- you might think first that "Jack Russell" is a name, and it is not "Jack Daniels" which would bring you high spirits :-)
Back to "Jack Russell Terrier"= a name + Terrier. The word Terrier activates the knowledge that this word represents a dog breed. In your mind might be activated the concept and image of a dog, little, white, energetic, barking, chasing a bunny, or a cat, dog food, toys, etc. Then you might remember that your teacher has a JRT named Taylor. And that brings you into mind that my last name is Taylor as well (if spelled in English). That multitude of items that were activated by the net of concepts and information linked to the stimulus “JRT” is your network and it was activated because your mind uses Parallel Distributed Processing (PDP).


PDP effect in information processing be it visual or auditive is that we fill in the gaps.
PDP is like the new style Christmas lights. If you are old enough you remember the old style ones when a bulb was going off the entire string was off.That is in electric terms a serial connection. Like a logical path, just like a computer has a logical programming and follows step-by-step a Yes_NO rule.
The new lights are still blinking even if one is off. That is a parallel connection. Our mind works in parallel, when one information is activated that also will activate neurons in brain that hold information in relation to the one activated. Like a fishing net. Lots of nods and links between the nods. If you pick up one nod in one place of the net it will come up with a certain area that has all the connections to the one nod.
If I say "Tiger," you will remember in general what I say: big cat, stripes, wild cat, jungle, dangerous, zoo, in extinction, Bengal Tiger, etc. Lots of information will cross your mind. It is all because of the PDP processing!

Face recognition takes into account the entire picture of a face not the separate items. The 3D theory of geons does not quite work in face recognition since we have a hard time recognizing separate items of a face (mouth, nose, eyes, taken separate are very hard to be recognized).

We recognize faces using the Gestalt theory, we see the entire picture, the whole before the parts.

Prosopagnosia= inability to recognize faces even if the person affected is able to still recognize objects. Problem in the inferotemporal cortex.