Pattern Recognition
Look at the letters printed below.
A a A a A a A
It is easy to see that all of the letters are “A”. Our cognitive processing system recognizes the patterns as “A” despite the fact that the letters differ in case (upper versus lower), font, print style (bold versus normal) and size. This example demonstrates the power of human pattern recognition. We are far more accurate and faster than computers at recognizing patterns as similar regardless of variations in size, coloring and shape.
Cognitive psychologists have grappled with our ability to recognize patterns in their many variations and have arrived at several theories to explain how we can do this with good accuracy.
- Template theories state that we carry around a series of patterns or templates in our long-term memory and we match incoming patterns against these templates. If there is a lot of similarity between the input pattern and the template pattern, then there is a match and we recognize the input as being an example of the template. Template theories would argue that we have a template for the letter “A” that is a pretty good match for all of the letters printed above.
- Feature theories state that we recognize patterns by their parts. For example, uppercase letters from the English alphabet share many features; E and F both have horizontal and vertical straight lines as features. Similarly, you may describe a friend’s face by a series of features…eye color, shape of nose and mouth and so forth.
- Distinctive features help us to distinguish one pattern from another. For example, an uppercase E differs from an uppercase F by the bottom horizontal line. Thus, this bottom line is a distinctive feature that helps us distinguish between the two letters.
Theories on Pattern Recognition Continued
3. Structural theories place emphasis on how features of a pattern are joined to other features of a pattern. These theories emphasize that we follow a set of rules that help us to recognize patterns.
- Structural theorists often use figures called ambiguous figures to demonstrate this rule principle.
Look at the figure below.
Figure 2.1
It can either be seen as a white vase on a blue background or as two blue profiles of people facing each other. This face/vase figure is called ambiguous because it can be seen in one of two ways.
Whether you see the figure as a face or a vase, you are using a rule called figure/ground to recognize a pattern. The figure/ground rule states that we tend to organize patterns so that there is something in the foreground, a figure that has a form, and something else in the background, which tends to be more formless and does not capture our attention. When you organize the face/vase figure as a white vase, then the blue area becomes the background upon which the white figure rests. When you organize the face/vase figure as two profiles, then the blue area is seen as a figure with the white background separating the two profiles from each other.
The Cognitive Processing System
Patterns tend to have many component parts. So whether you are a template, feature or structural theorist, you have to ask the question about how the cognitive processing system goes about analyzing these various parts.
- Serial processing is processing that carries out one operation at a time. In other words, you process a pattern one feature at a time.
- Parallel processing is processing that carries out more than one operation at a time. In other words, you process many of the features of a pattern at the same time.
- The context in which we view a pattern is also important. For example, we can identify a letter more rapidly if it is part of a meaningful word when compared to our speed of identifying a letter presented alone or in a non word. This is called the word superiority effect and it is a good example of top-down processing. Knowledge about the words of a language stored in long-term memory interact with incoming letter features to give information about what letters are in a word. This results in faster times for recognizing the letter.
- The interactive activation model that tries to explain the word superiority effect is diagrammed below. The fat arrows indicate excitatory connections between the levels of processing. Excitatory connections give positive evidence for the identity of a letter or word. The narrow arrows are inhibitory connections that provide negative evidence for the presence of letter or word. For example, a straight vertical line feature would be positive evidence for a number of uppercase English letters but not for others.
- The presence of a vertical line indicates that some letters are possible while others are not. The impossible ones are inhibited while the possible ones move on to the next level. At the word level, knowledge of where in a word a particular letter is likely to fall (beginning, middle, end) also inhibits some letters from the mix, leaving others that are more likely. The interaction between our knowledge of words and the incoming features gives rise to faster letter recognition when letters are presented in the context of meaningful words.
- In the diagram below, notice that the arrows point in one direction between the feature and the letter levels while they point in two directions between the letter and the word levels. The two-directional arrows describe the top-down processing component of this model whereby our knowledge of words and the placement of letters within those words is influencing the speed with which we can recognize individual letter features.
Figure 2.1
This concludes the lesson 2 commentary please complete the Practice Quiz before moving onto lesson 3.