Free Web Hosting Provider - Web Hosting - E-commerce - High Speed Internet - Free Web Page
Search the Web

Memory and Illusions    Framing Effects    Prospect Theory   Fuzzy-trace theory  Discussion

Go to Previous PageHome PageGo to Next Page

Information theorists used memory variations to explain errors in decision-making, and performance on many difficult cognitive tasks (Reyna & Brainerd, 1995). The historical background of information theory can be traced back to cognitive development theories.  Research in cognitive development focused on the emergence of analytic and logical thought, as reflected in theories of conscious experience (Anooshian, 1998). Traditional theorists like piagetians and constructivists believed that reasoning shaped memory and that memory was subordinated to reasoning (Reyna & Brainerd, 1995).  They focused on non-memorial explanations for reasoning while acknowledging the necessity of memory.  In other terms, memory was necessary but not sufficient for reasoning.  The following statement exemplifies this conclusion.

The schemata used by the memory are borrowed from the intelligence.
- Piaget & Inhelder, 1973 (Reyna & Brainerd, 1995, page 3)


Unlike constructivists and piagetian theorists, information-processing theorists viewed memory as crucial for reasoning (Healy & McNamara, 1996). They characterized reasoning almost in terms of the ways in which a computer processes information.  Information is inputted, computations occur, and information is elicited (Reyna & Brainerd, 1995).  Without memory, reasoning would not take place.  They believed this relationship between memory and reasoning to be a delicate one.  Most information-processing explanations of reasoning revolved around the capacity-limitations of short-term (or working) memory (Cox, 1980; Garner, 1970; Jou et. al., 1996; Miller, 1956; Reyna & Brainerd, 1995).  They believed that these limitations put heavy constraints on both memory and reasoning.   Information processing theorists discussed the organized knowledge structures stored in memory that are used to guide comprehension and memory.  Miller (1956) provided an excellent discussion on information measurement and how we can interpret this theory in psychological terms.  Briefly, information theory can help us understand how information can be seen as variance, and the amount of transmitted information as either correlation or covariance (see Figure 1).  Therefore, output will depend on the input, or will be correlated with the input (Miller, 1956).  Thus, the measure of transmitted information is simply the input-output correlation.

Figure 1.  Input-Output Correlation Diagram
Information Correlation of Input Output Data
Source: Stephanie Barclay McKeown, Interpretation of George Miller's 1956 article.

Garner (1970) argued that too much emphasis has been placed on how we process information, but little has been placed on the nature of the stimulus.  Without the input, no processing would ever take place.  He stressed the need to consider how the stimulus should be defined in information processing as an empirical question that could not be answered a priori.  He argued that we do not make generalizations from a study with only one subject, so we should not make generalizations about one stimulus or one task, but we should investigate the roles of many stimuli and a multitude of tasks.  He stressed that researchers have also not given sufficient consideration in studies of information processing that humans are capable of many different types of information processing.  He asked why so many asked the question whether humans process information linearly (in serial) or in parallel.  One of the assumptions asserted in Reyna and Brainerd’s understanding of memory was that experiences are encoded roughly in parallel (Reyna & Brainerd, 1995).  Garner (1970) thought possibly we could do both depending on the task and the stimuli – or perhaps we could do them simultaneously.  Similarly to Reyna & Brainerd, he believed that stimuli are not all equivalent, and all information cannot be processed in the same way.  We shouldn’t be seeking a single solution but a diversity of answers to understanding stimuli in information processing.  He stressed that we must be prepared to ask when and under what circumstances do we process information in a particular way.

From an information processing perspective, if the question posed to the respondent and type of response alternatives are accurate, then nearly all the input information will be transmitted and will be recoverable from the responses.  However, if the information is not accurate, if the information provided is inadequate or ambiguous, then the transmitted information will be considerably less than we require when making accurate inferences about how people process information.  Further to this, if the response options available to the participant do not fully satisfy the respondent’s decision choice, we gain little knowledge about decision making.  In line with Garner’s view of investigating different types of stimuli are researchers who have investigated different levels of information provided to the respondent in a decision making scenario (Kuhberger, 1995; Levin et. al., 1985; Levin et. al., 1986).

The experimental problem is to increase the amount of input information, and to measure the amount of transmitted information.  However, through information processing, when we increase the input information the transmitted information also increases but eventually levels off at a particular value (Miller, 1956).  This value is the channel capacity.  The channel capacity of the observer represents the greatest amount of information that can be given about the stimulus on the basis of absolute judgment.  In terms of stimulus-centered framing problems, channel capacity is the greatest amount of information that can be conveyed accurately through the provision of the appropriate number of response alternatives.  As mentioned previously, if the information provided to the participant is inadequate or ambiguous, the respondent will be unable to understand the problem, and will be unable to provide accurate information on the basis of the available response options.   If the channel capacity is maximized, then measurement error could occur.  The difficulty with providing the appropriate information can be seen as noise.  Through investigation and careful question and response construction, noise can be decreased to an acceptable level determined by the researcher (see Figure 2).

Figure 2.  Diagram of Information Processing Theory

Source:  EPSE 501 Class Web site WebCT

The channel capacity is the upper limit on the extent to which the respondent can match his/her responses to the response alternatives provided (Cox, 1980).  If information is not transmitted appropriately due to the maximized channel capacity (measurement error) the data obtained from any analysis using that particular problem formulation might not be useful for any inferences at all.  The limitations of information theory are that it is imprecise and one-dimensional.  According to information theory, the optimal number of response categories is very much connected to the number of stimuli being scaled.  In other words, if only one stimulus was being scaled, only one response category would be needed (absolute judgment).  However, these options may not provide the respondent with the natural response he/she may wish to select.  Also, information theory supports the notion that once the channel capacity is maximized the respondent will not process any more information (Cox, 1980; Miller, 1956).  A way around overburdening the channel capacity is to increase the dimensionality of the stimulus.  The notion of increasing dimensionality of the stimuli is presented through discussion of the absolute judgment paradigm.

In his discussion of absolute judgment paradigm, Garner (1970) discussed how these were used frequently with multidimensional stimuli.  The primary experimental problem studied with this technique has been to ask whether information transmission is increased when new stimulus dimensions are added to an existing one.  The experimental literature reviewed by Garner showed that when dimensions are added orthogonally, and judgment of all dimensions is required, performance improves, but not as much as would be expected if the subject were able to deal with each dimension (also see Miller, 1956).  On the other hand, if a new dimension is added redundantly by being correlated with an existing dimension, there sometimes is gain in performance, but in other cases there is little or no gain.  However, redundancy should not be seen as negative.  It is useful in increasing the amount of information transmitted to the respondent (Garner, 1970).  Absolute judgment is limited by the amount of information provided to the respondent.

1.  Gain the students' attention.
  • Use cues to signal when you are ready to begin.
  • Move around the room and use voice inflections.
  • 2.  Bring to mind relevant prior learning.
    • Review previous day's lesson.
    • Have a discussion about previously covered content.
    3.  Point out important information.
    • Provide handouts.
    • Write on the board or use transparencies.
    4.  Present information in an organized manner.
    • Show a logical sequence to concepts and skills.
    • Go from simple to complex when presenting new material.
    5.  Show student how to categorize (chunk) related information.
    • Present information in categories.
    • Teach inductive reasoning.
    6.  Provide opportunities for students to elaborate on new information.
    • Connect new information to something already known.
    • Look for similarities and differences among concepts.
    7.  Show students how to use coding when memorizing lists.
    • Make up silly sentences with first letter of each word in the list.
    • Use mental imagery techniques such as the keyword method.
    8.  Provide for repetition of learning.
    • State important principles several times in different ways during the presentation of information (STM).
    • Have items on each days's lesson from preivous less (LTM). 
    • Schedule periodic reviews of previously learned concepts and skills (LTM).
    9.  Provide opportunities for overlearning of fundamental concepts and skills.
    • Use daily drills for arithmetic facts.
    • Play form of trivial pursuit with content related to class.