Humans as Information

Processors
Information systems are designed, of course, to be used by us humans. It's important, therefore,
to understand generally how most of us process and interpret information.

INTRODUCTION

For many if not most systems, interfaces with human users likely represent the most success-
critical components. The interface, to the user, is the system. Management information and
decision support systems require regular interaction between the system and decision makers, its
users. The information system design, especially with respect to its user interface, often dictates
how a great deal of the work of an organization will be performed, the order of task performance,
and the importance of individual tasks. Ultimately, the effectiveness of the interface in
structuring, facilitating, and supporting the work of the organization will affect the achievement
of organizational objectives. To understand good system design characteristics, it is important to
understand how humans process information.

Two model frameworks, a general model and the Newell-Simon model of the human as an
information processor, may be helpful in understanding the concept. Tentative limits on human
information processing capabilities and the relationship between information and performance,
along with related concepts and research from other fields, may provide additional insight.

HUMAN AS INFORMATION PROCESSOR - GENERIC MODEL

A human as an information processor may be represented by a simple model in which the senses
(sight, hearing, etc.) are receptors that pick up signals and transmit them to the brain (processing
unit with storage). After processing, the person produces output responses (physical actions,
speech, etc.).
A human's ability to accept inputs and produce responses is limited; when the capacity is
exceeded, information overload may negatively affect response and performance.

A person often receives more input than s/he is capable of accepting and processing, resulting in
a need to manage the quantity of input to prevent information overload. A common way to do
this is to use a filtering or selection process to block some inputs. Typical filtering may be based
on the person's frame of reference (based on experience and knowledge), normal decision
procedures, or stress in the decision situation. Deadlines, for example, may be stressful and force
a manager to focus on only the most important decision inputs, filtering out lesser ones.

Frame of reference filtering occurs with both input and processing; using experience often
consists of using filters that have worked well in the past. The brain, over time, categorizes and
patterns data, which becomes a part of how the person develops understanding of an event or
situation. These patterns or frames of reference then help reduce processing and input
requirements. Effective use of frames of reference accumulated over a long period of time is an
element of expertise.
Data inconsistent with a frame of reference are blocked by the filtering process. In combination
with human input receptor limitations, this may lead to perceptual errors (e.g., distortions,
inferences, and omissions) which reduce information by increasing uncertainty. Anyone who
has ever had a message misunderstood can relate to this issue.

Allen Newell and Herbert A. Simon (1972) developed a human problem solving model. In
effect, this particular model uses a computer problem-solving analogy. The figure below
compares the Newell-Simon model of information processing with a general model of a
computer system.

The Newell-Simon Model of humans as information processors

The components of a human information processing system are a processor, sensory input, motor
output, and three different memories: long-term memory (LTM), short-term memory (STM), and
external memory (EM). Humans process information serially, only one task at a time.
Computers, on the other hand, may use either serial or parallel (more than one task at a time)
processing.

Operation Serial Design Parallel Design

Data transferred a word at a time (e.g., 32
Data transfer Data transferred a bit at a time.
bits).
One adder-comparator. One adder-comparator for each set of bits.
Arithmetic on one
Operations on pairs of bits from Operations on all pairs of bits
data item
right to left. simultaneously.
Processing of One processor. Processing of CPU consists of several small processors.
several data items one item at a time. Several items processed concurrently.

How can humans work on more than one task at a time, as many apparently do? They probably
accomplish it by rapidly switching from one quick task to another. Humans may also use pattern
matching. Have you ever read a job ad that said "Must be capable of multitasking." ?

In the Newell-Simon model, the long-term memory has essentially unlimited capacity. Its
content consists of symbols and structures of chunks, or units of stored information. Short-term
memory is very small, capable of holding only a few symbols, and is part of the processor. Quick
read and write times are characteristic of short term memory. The computer analogy is RAM
registers which temporarily hold data. External memory involves external media (notebook,
chalkboard, etc.); it compensates for limited short term memory and is fairly efficient to bridge
long and short term memory.
In the Newell-Simon model, the task environment is the problem as it is presented; the problem
space is the way a particular problem solver defines the task. In other words, confronted by a
problem, the problem solver develops a model to use in working on the problem, or a conceptual
"space" where problem solving takes place. Thus the problem space, while necessarily related to
the task environment, is not identical to it.

Problem solving thus consists of developing a problem space from the task environment and then
performing processing operations within the problem space until a solution is found. The
structure in the problem space, as defined by the problem solver, includes some redundancy,
helping predict aspects of another part of the problem space; this generally makes for a more
efficient process.

As previously discussed, many decisions are made in accordance with the descriptive
(satisficing) model rather than the normative (optimizing) model. Instead of always using an
objective or algorithmic reasoning process, many decision makers utilize judgmental rules of
thumb known as heuristics which simplify the search process by eliminating alternatives without
explaining them. The use of heuristics can be both efficient and effective if the decision maker
has sufficient experience and judgment. It is also true, unfortunately, that heuristics are often
used in ways that result in poor decisions.

A related concept is that of bounded rationality. Because of normal human limitations on rational
thinking, we often simplify a problem in order to be able to understand and deal with it; even
though our decisions with respect to the simplified problem may be rational, it is not necessarily
true that the same decisions are rational with respect to the real problem. In addition to human
limitations on processing capacities, bounded rationality also results from individual differences
in such factors as age, education, cultural background, and attitude. One characteristic of a
successful problem solver is adequate ability to define problem space (i.e., set boundaries).

Human limitations, especially as related to heuristics, includes short term memory limitations,
inability to detect differences, and inability to deal with probabilistic data. Miller (1956)
suggested "the magical number seven, plus or minus two" as describing the limitation on short-
term memory. His results and others indicate that humans can generally hold only five to nine
symbols or "chunks" of information in short-term memory. These limits are especially important
with respect to human processing of codes, quantities, and other single-symbol data. Graphic
information, "pre-chunked" as it is, can permit humans to process more information at a time.

The tasks of those who use information often include error detection (noticing differences
between incorrect and correct data) and reacting to variations in data. According to Weber's law,
the difference noticeable by most humans is a constant proportion (or percentage) of the physical
dimensions of a stimulus. In other words, as the dimensions change, the amount of change
required to be noticeable also changes (to stay a constant percentage of the dimension). If a
person notices a 1-pound increase in a 5-pound weight, s/he will notice a 2-pound increase in a
10-pound weight. The difference required to distinguish among heavier objects is larger, but the
relative or proportional difference is the same. Research results indicate that Weber's law holds
true for processing data. A variation from total revenues of $100,000 on budget of $1,000,000
appears to have the same "noticeability" (10%) as a variation in indirect costs of $4,700 on a
budget of $47,000. An everyday example is the lack of public acceptance of the Anthony dollar;
although many countries have similar-sized coins of differing denominations, its size was just
too close to that of a quarter for us to use it.

According to Wright (1980) and others, humans do not have good intuition in assessing
probabilities of various events. Decision makers are often called upon to identify correlation and
causality; people often observe dependencies between two variables and conclude erroneously
that there is an association when in fact there is not. A number of biases may cause errors in
estimating probabilities. These may be based on availability (events easily remembered or
imagined are assigned higher probabilities), recency (recent results given greater weight), and/or
hindsight (people who are told that an outcome has happened in the past give it a higher
probability than those without the information).

Concreteness and anchoring and adjustment are examples of strategies or biases adopted by
people to help deal with their processing limitations. Concreteness means that a decision maker
tends to use information that is readily available and only in the form in which it is presented,
and tends not to search stored data or manipulate data presented. For example, suppose you
supervise an employee whose actual job performance is difficult to objectively measure. What
you do (concretely) know is that the employee in question has an erratic attendance record, has
received a disciplinary warning, and was fired from a previous job. Using concreteness, you are
likely to give the known factors greater weight than they deserve because you have little
information on more important aspects of the person's work. Teachers write words on a
chalkboard or project images on a screen because doing so make them more concrete to a class.

Anchoring and adjustment means that individuals often make judgments by establishing an
anchor point and making adjustments from this point, thus reducing information processing
requirements. Though often used in such activities as pricing and budgeting, it has the problems
of overemphasizing recent results or decisions (thus underemphasizing new results or ideas) and
overstating the value of the anchor point, which may have been chosen using inappropriate, or
arbitrary criteria. Even if it was appropriate when chosen, environmental conditions on which it
was based may have changed such that it is no longer useful. If last year's advertising budget was
based on optimism and this year's is based on a percentage increase, it is not necessarily true that
last year's optimism should or does still prevail.

COGNITION

An area of psychological research, human cognition, helps explain how information systems can
improve human capabilities. Cognition refers to "the activities by which an individual resolves
differences between an internalized view of the environment and what is actually perceived in
that same environment." (Zmud, 1979) Cognitive models are representations of cognitive
processes (e.g., the Newell-Simon (1972) model, Festinger's (1957) Theory of Cognitive
Dissonance).
Cognitive style is a term used for the the process by which humans organize and modify
information as they make decisions. McKinney and Keen (1974) suggested two dimensions -
information gathering (organization of stimuli) and information evaluation (how information is
analyzed). Individuals ware classified on the information gathering dimension along a continuum
ranging form perceptive (generalization is made from relationships among data items) to
receptive (specific knowledge is developed from details in the data). Classification on the
information evaluation dimension is based on a continuum ranging from systematic (analytic) to
intuitive (heuristic). A systematic decision maker uses structure and deduces conclusions to solve
problems, while an intuitive decision maker uses trial and error or reacts to new information.

Are you a visualizer or a verbalizer?

Since most information system designers are often systematic in their approaches to decision
making, managers who are also systematic will find satisfaction using such a system. For a
heuristic decision maker, on the other hand, an effective information system must be
different, allowing for more alternatives, allow modification of processing order, and offer
options with respect to level of detail presented and form of output. Since many individuals,
however, are both systematic and heuristic as well as adaptable, the importance of this issue may
not be great. A well-known humorous (though perhaps factual) example is that women read
maps and ask for directions (systematic) while men drive until they find the right place, often
getting lost!

Neuropsychology suggests that biological differences in brain hemisphere dominance may

explain different approaches to problem solving (e.g., Robey and Taggart, 1982), with the left
side of the brain used for rational or analytic processing and the right side for intuitive or creative
processing. While each of us should be good at both types (assuming both halves of our brain are
equally effective), we each have a different set of experiences which lead us to prefer to use one
half of the brain over the other. We do, however, tend to "use what we have," and therefore we
should be able to adapt to an information system that does not accommodate our processing
preference. But if a person performs better using one type of processing, failure to provide an
information system that allows the person to use his or her strength may hinder organizational
goal achievement and/or result in inaccurate performance evaluation.

HCI: HUMAN-COMPUTER INTERACTION RESEARCH

The relatively new and rapidly developing field of human-computer interaction studies numerous
aspects of hardware and software design as they relate to the human-computer interface,
especially in the workplace. Some samples of research and academic programs at a number of
universities include those at Cornell, Uppsala (Sweden), Carnegie Mellon, Binghamton, and
Stanford. A brief history of the field may be examined here (from Carnegie-Mellon). HCI, as it's
called, has in addition to degree programs its own journal, a number of institutes, numerous
laboratories, and library databases. HCI is much broader than the scope of this course, but some
aspects of it are quite relevant to the study of management information systems and technology
management.
 EXAMPLE HCI RESEARCH TOPIC: SEARCHING A LIST OF INFORMATION

Decision makers and other knowledge workers often must find a reference or other item by
searching a long list. A number of computer search algorithms are available which are extremely
efficient. For purposes of this discussion, however, the focus is on the interface with which
users interact with information. An interesting demonstration comparison of the "fisheye"
approach (developed at the University of Maryland) with traditional approaches (arrow bar,
scrollbar, and hierarchy) may be tested at their website here. Which is faster? Which do you
prefer?

OTHER CHARACTERISTICS

Humans have a psychological need for feedback to be assured that output was received or that
input was accepted. System design considerations for system feedback response include data
entry speed and accuracy. In addition, people seem to attach a psychological value to unused
data; it's very common for today's managers to accumulate and store of a great deal of data that
will never be used. This can be uneconomical for several reasons. Decision makers seem to have
greater confidence when they have "extra" information and to see a symbolic value to having lots
of information at their disposal (i.e., it symbolizes rational decision making).

The previously mentioned problem of information overload has been made worse by dramatic
reductions in costs with concurrent increases in processing speed, storage capacity, and
communications capability. Where possible, systems should be designed to reduce the quantity
of information (e.g., by filtering and summarizing) to counteract managers' natural inclination to
prefer and to accumulate more information when they can do so. Studies have shown that
managers given summarized information generally make better and more consistent decisions,
but tend to be less confident in them.

Individual differences affect how decision makers use systems as well as how satisfied they are
with system performance. Ideally, systems should be designed to accommodate these
differences. As an economic and practical matter, however, individually tailored systems are not
always possible. Some of these differences, as documented in research findings, are provided
below.

Differences Relationship with Information Processing

Locus of control (internal-external) Internal locus of control: more search activity than external (Lefcourt, 1972; Phares, 1976).

Low dogmatism: more search activity, more deliberation, and less confidence in decisions (Lambert & Durand,
Dogmatism (low-high)
1977; Long & Ziller, 1965; Taylor & Dunnette, 1974).

Risk-taking propensity (low-high) High risk-taking propensity: more search activity than low (Taylor & Dunnette, 1974).

Extroverts: quicker retrieval from long-term memory, better short-term retention, and less long-term retention
Extroversion-introversion
than introverts (Eysenick, 1977).
 Lower tolerance for ambiguity: preference for concrete information, perception that more information will be
Tolerance for ambiguity (low-high)
valuable (Dermer, 1973).

High intelligence: faster information processing, faster decisions, better information selection, better retention,
Intelligence (low-high)
better internal organization of information (Taylor & Dunnette, 1974; Hunt & Lansman, 1975).

High quantitative abilities: more use of short- term memory, less use of long-term memory (Hunt & Lansman,
Quantitative abilities (low-high)
1975).

Verbal abilities (low-high) High verbal abilities: better short-term memory (Hunt, Frost, and Lunnebourg, 1973).

Experience: better information selection, less effective integration, greater flexibility, and less confidence
Experience in decision making
(Taylor & Dunnette, 1974).

Task knowledge (low-high) High task knowledge: less information search (Benbasat & Schroeder, 1978).

Older subjects: more information search, better information selection, more flexibility, and slower decision
Age
making (Taylor, 1975; Taylor & Dunnette, 1974; Eysenck, 1977).

Management Level High management level: quicker decisions (Taylor, 1975).

The organizational position of a person may have a significant effect on performance as an

information processor. This section will examine information processing by one organizational
position or role--a manager.

A study by Henry Mintzberg (1973) of managers in their jobs characterizes managerial work as
follows:

• Much work at an unrelenting pace,

• Activity characterized by brevity, variety and fragmentation,
• Preference for live action,
• Attraction to the verbal media,
• Network of contacts, and
• The manager's job is a blend of rights and duties.

Many of the decisions themselves are unstructured, and the environment in which decisions are
made may be unstructured as well. Some characteristics of systems used by managers which
should be incorporated where possible include:

• easily interruptible
• provide information that is easily combined with other information
• integration of multiple functions (e.g., spreadsheet analysis, word processing, and
Internet access)
IS DESIGN IMPLICATIONS

Filtering: Information systems should be designed to filter irrelevant data and to provide
increased filtering for stress decisions.

Systems: should attempt to override undesirable frame-of-reference filters by reinforced display

of relevant data.

Newell-Simon model: Information systems should assist in defining problem space and in the
search process for a solution. The information format should attempt to expand the limits of
bounded rationality. Systems should utilize the user memory that is suited to the task.

Magical number 7 + 2: Codes for human use should not exceed five to seven symbols or be
divided into segments of five or less. Systems should not have humans do significant, unaided
processing. Graphics may be used to present "chunks" of data in an efficient way.

Just noticeable differences: Systems should highlight significant differences rather than
assuming humans will notice them.

Humans as intuitive statisticians: The information system should provide statistical analysis of
data: sample variance, correlation, probability estimates, etc. Decision algorithms should provide
a consistency check

Bias of various information sources: Data generation procedures should be designed to assist in
eliminating bias such as recency of events.

Concreteness: The information needed should be presented in the form needed. No added
processing should be required.

Anchoring and adjustment: Information and decision systems should be designed to assist in
selecting a suitable anchor point and for prompting adequate adjustments from it.

Cognitive style: Where possible, systems should allow selection of alternatives for order of right
brain-left brain processing and forms of information presentation in order to accommodate
different styles.

Feedback: Systems should provide feedback to indicate that data has been received, processing
is taking place, etc. Response times should be such that throughput is meaningful and errors are
minimized.

Value of unused data: Explains some of pressure for data with no apparent utility. Suggests
storage and retrieval strategies and terminal access to increase availability without individual
storage.

Information overload: Input should be kept below the overload point. System use should not
involve managing or processing amounts of data beyond overload.
Individual differences: Those which are critical to system use should be identified and
differences explicitly accommodated, whenever possible, through a flexible interface.

Processing timing: Managers need short bursts of information processing to support their mode
of operation.

Amount of information: Information systems should present summarized data in a compressed

decision-impelling format, but the system should also allow browsing through the raw data.

REFERENCES

Ashton, R.: Human Information Processing in Accounting, Studies in Accounting Research,

American Accounting Association, 1982.

Benbasat, Izak, and R. G. Schroeder: "An Experimental Investigation of Some MIS Design
Variables," MIS Quarterly, 2:2, 1978, pp. 43-54.

Conrad, R.: "Errors of Immediate Memory," The British Journal of Psychology, November1959,
pp. 349-359.

Crannell, C. W., and J. M. Parrish: "A Comparison of Immediate Memory Span for Digits,
Letters and Words," The Journal of Psychology, 44, October 1957, pp. 319-327.

Dermer, J D.: "Cognitive Characteristics and the Perceived Importance of Information," The
Accounting Review, 48, 1973, pp. 511-519.

Dickson, G. W., J. A. Senn, and N. L. Chervany: "Research in Management Information

Systems: The Minnesota Experiments," Management Science, 23:9, May 1977, pp. 913-923.

Eysenck, M. W.: "Human Memory: Theory, Research and Differences," Pergamon, Oxford,
1977.

Gul, Ferdinand A.: "The Joint and Moderating Role of Personality and Cognitive Style on
Decision Making," The Accounting Review, 59:2, April 1984, pp. 264-277.

Huber, George P.: "Cognitive Style as a Basis for MIS and DSS Designs: Much Ado About
Nothing?" Management Science, 29:5, May 1983, pp. 567-597.

Hunt, E., N. Frost, and C. Lunnebourg: "Individual Differences in Cognition: A New Approach
to Intelligence," in G. H. Bower (ed.), The Psychology of Learning and Memory, Academic
Press, New York, 1973.
Hunt, E., and M. Lansman: "Cognitive Theory Applied to Individual Differences," in W. H.
Estes (ed.), Handbook of Learning and Cognitive Processes, Earlbaum, Hillsdale, NJ, 1975.

Jenkins, A. Milton, and Randall D. Johnson: "What the Information Analyst Should Know About
Body Language," MIS Quarterly, 1:3, September 1977, pp. 33-47.

Lambert, Z. V., and R.M. Durand: "Purchase Information Acquisition and Cognitive Style,"
Journal of Psychology, 97, 1977, pp. 3-13.

Lefcourt, H. M.: "Recent Developments in the Study of Locus of Control," in B. A. Maher ed),
Progress in Experimental Psychological Research, Academic Press, New York, 1972.

Libby, R.: Accounting and Human Information Processing: Theory and Applications, Prentice-
Hall, Englewood Cliffs, NJ, 1981.

Long, B. H., and R. Z. Ziller: "Dogmatism and Predecision Information Search," Journal of
Applied Psychology, 49, 1965, pp. 376-378.

McKenney, James L., and Peter G. W. Keen: "How Managers' Minds Work," Harvard Business
Review, May-June 1974, pp. 79-90.

Miller, George A.: "The Magical Number Seven, Plus or Minus Two: Some Limits on Our
Capability for Processing Information," The Psychological Review, 63:2, March 1956, pp. 81-97.

Mintzberg, Henry: The Nature of Managerial Work, Harper & Row, New York, 1973.

Newell, Allen, and Herbert A. Simon: Human Problem Solving, Prentice-Hall, Englewood
Cliffs, NJ, 1972.

Phares, B. J.: Locus of Control in Personality, General Learning Press, Morristown, NJ, 1976.

Robey, D., and W. Taggart: "Human Information Processing in Information and Decision
Support Systems," MIS Quarterly, 6:2, June 1982, pp. 61-73.

Shneiderman, Ben: Software Psychology: Human Factors in Computer and Information Systems,
Winthrop, Cambridge, MA, 1980.

Simon, Herbert A., and Allen Newell: "Human Problem Solving: The State of the Theory in
1970," American Psychologist, 26, February 1971, pp. 145-159.

Slovic, Paul: "From Shakspeare to Simon: Speculations--and Some Evidence--about Man's

Ability to Process Information," Research Monograph, 12:12, Oregon Research Institute,
University of Oregon, April 1972.

Taylor, R. N.: "Age and Experience as Determinants of Managerial Information Processing and
Decision Making Performance," Academy of Management Journal, 18, 1975, pp. 74-81.
Taylor, R. N., and M. D. Dunnette: "Relative Contribution of Decision-maker Attributes to
Decision Process," Organizational Behavior and Human Performance, 12, 1974, pp. 286-298.

Tversky, A., and D. Kahneman: "The Framing of Decisions and the Psychology of Choice,"
Science, 211, January 1981, pp. 453-458.

Zmud, Robert W.: "Individual Differences and MIS Success: A Review of the Empirical
Literature," Management Science, 25, October 1979, pp. 966-979.

Portions adapted from Management Information Systems: Conceptual Foundations, Structure,

and Development, Second Edition, by Gordon B. Davis and Margrethe H. Olson (New York:
McGraw-Hill, Inc., 1985).