Semantic feature-comparison model

Semantic Feature Comparison Model is used “to derive predictions about categorization times in a situation where a subject must rapidly decide whether a test item is a member of a particular target category” [1]. When pressed with time, how do we make judgments and place items or events in particular categories? That is what the model tries to answer. As such, in this semantic model, there is an assumption that certain occurrences are categorized using its features or attributes of the two subjects that represent the part and the group. For example, which is often used to explain this model, the statement a ‘robin is a bird’. The meaning of the words robin and bird are stored in the memory by virtue of a list of features which can be used to ultimately define their categories, although the extent of their association with a particular category varies.

History

This model was conceptualized by Edward Smith, Edward Shoben and Lance Rips in 1974 after they derived various observations from semantic verification experiments conducted at the time. The task is simple: respondents merely have to answer ‘true’ or ‘false’ to given sentences. Out of these experiments, they observed that people respond faster when (1) statements are true, (2) nouns are members of smaller categories, (3) items are ‘typical’ or commonly associated with the category (also called prototypes), and (4) items are primed by a similar item previously given (University of Alaska Anchorage, n.d.), such as previous statement ‘eagle is a bird’ and next statement ‘robin is a bird’. In the latter item, respondents will respond faster to the latter statement since the category bird has been primed. So, based on the previous observations, the proponents were able to come up with the Semantic Feature Comparison Model.[1]

Theory

The cognitive approach consists of two concepts: information processing depends on internal representations, and that mental representations undergo transformations. For the first concept, we could describe an object in a number of ways, with drawings, equations, or verbal descriptions, but it is up to the recipient to have a background understanding of the context to which the object is being described in order to fully comprehend the deliverable. The second concept explains how memory can alter the way we perceive representations of something, by determining the sequence in which the information is processed based on previous experiences.

Features

The main features of the model, as discussed by Smith et al (1974), are the defining features and the characteristic features. Defining features refer to the characteristics that are essential elements of the category, the non-negotiable, so to speak. For example, the ‘bird’ category includes such defining features as ‘they have wings,’ ‘feathers,’ ‘they lay eggs,’ etc. Meantime, characteristic features refer to the elements usually found or inherent to category members but are not found in all, or non-essentials. For example, birds ‘fly,’ – that is characteristic because while most birds fly, there are some who cannot. The model has two stages for decision making. First, all features of the two concepts (bird and robin, in our example) are compared to find out how alike they are. If the decision is that they are very similar or very dissimilar, then a true or false decision can be made. Second, if the characteristics/features are in-between then the focus shifts to the defining features in order to decide if the example possesses enough features of the category, thus, categorization depends on similarity and not on the size of the category.

Experiments

Structure and Process in Semantic Memory: A Featural Model for Semantic Decisions (Edward E. Smith, Edward J. Shoben, and Lance J. Rips, Stanford University

Experiments

Two separate experiments were conducted to support the model proposed by Smith, Shoben, & Rips, 1974. The first experiment tested category size in determining semantic decisions and the second experiment was called the Instance-Category Verification Experiment. To begin, the first experiment, the statement “An S is a P” is typed onto two different cards. One card for the instance and smaller category and the other card are for the instance and the larger category. All 52 true statements are randomized with 52 false statements by pairing instances and categories from different triples. The instance-category pairs were then presented individually to a subject for rapid verification. In the second experiment, subjects are presented a set of instance-category pairs (Several instances were used for each category) and then were asked to rate the typicality of each instance to its associated category. In the both experiments, the Featural Model for Semantic Representation quantifies its findings by making the value, x, is equal to the overall similarity between the instance and the target category. The x value determines whether the participants execute a fast reaction time in both true and false response. True responses are (x>c1) and False responses are (x<c1). When the participants are having a difficult time determining the similarity between the instance and the target category, (co<x<c1), the second stage of the semantic feature model is entered; which focuses more on the defining features of the instance and the target category. Four expressions where derived to quantify the probability of each error type and to express correct true and correct false reaction times. 30 young adults from Stanford University were the subjects in both studies.[1]

Michael Posner Experiment (1986 at the University of Oregon)

Experiment

Subjects where shown two letters and asked to distinguish if the letters were the same or different.

Same meaning that the letters were either:

1. Identical (AA)
2. The Same Letter (Aa)
3. Both Vowels (AU)
4. Both Consonants (BD)

Different meaning that the letters were:

1. Different Categories (AD)

Purpose

The experiment was performed to help prove the hypothesis that multiple representations of stimuli.

Representations

One representation is based on the physical aspects of the stimulus, the visual derived representation of the shape presented on the screen. The Second representation reflects the fact that many stimuli can correspond to the same letter, for example recognizing different cases and fonts of a letter as the same letter. Thirdly abstraction representation represents the category the letter belongs to, vowel or consonant.

Result

It can be inferred from the experiment that the physical representations are activated first, phonetic representations next, and category representations last.

Saul Sternberg Experiment (1975 at Bell Laboratories)

Experiment

A subject was shown a set of letters and subsequently shown a single letter and asked if the single was part of the original letter group. After the subject successfully identified if the single letter was part of the group the subject was shown a larger set of letters. The subject was then asked to identify if a single letter was part of the second (larger) set. The subject went through this process 4 times.

Hypothesis

To respond to this task 4 primary mental operations must take place:

1. Encode: The Subject must identify the visible target.
2. Compare: The subject must compare the mental representation of the target with representations of the items in memory.
3. Decide: The subject must decide whether the target matches one of the memorized items.
4. Respond: The subject must respond appropriately for the decision made in step 3.

Basic Question

How to characterize the efficiency of recognition memory.

1. Parallel- the reaction time should be independent of the number of items in the memory set.
2. Serial-The reaction time should slow down as the memory set becomes larger, because more time is required to compare an item with a large memory list than a small memory list.

Result

The experiment convincingly support the serial hypothesis. As more letters are added to the set the reaction time for both no and yes responses increases linearly.

References

[1]<Smith, E. E., Shoben. E. J., and Rips, L. J. (1974). Structure and Process in Semantic Memory: A Feature Model for Semantic Decisions. Psychological Review, 81(3), 214-241>

[2]<University of Alaska Anchorage (n.d.). Cognitive Psychology - Memory Models, Knowledge Representation. Retrieved November 5, 2012 from http://www.math.uaa.alaska.edu/~afkjm/cs405/handouts/psycho.pdf>

[3]<Gazzaniga, Michael S., Richard B. Ivry, and G. R. Mangun. "Methods of Cognitive Neuroscience." Cognitive neuroscience: the biology of the mind. Third ed. New York: W.W. Norton, 1998. 111-112. Print.>

[4]<Gazzaniga, Michael S., Richard B. Ivry, and G. R. Mangun. "Methods of Cognitive Neuroscience." Cognitive neuroscience: the biology of the mind. Third ed. New York: W.W. Norton, 1998. 112-114. Print.>