Ch. 6— Alternatives to Animal Use in Research • 137 
pies of recent attempts toward computer simula- 
tion in behavioral research are listed in table 6-4. 
Computer simulations are used in behavioral re- 
search in a number of ways. Statistical simulations, 
in particular, are increasingly frequent. For ex- 
ample, one computer program simulates random- 
choice behavior in mazes (194), and two programs 
simulate random movements of animals under 
various conditions (17,49). The output generated 
by these kinds of simulations is compared with 
animal-generated data to see if factors other than 
pure chance are influencing the animals’ behavior. 
Statistical simulations are also used to test hy- 
potheses that may not be subject to empirical con- 
firmation. One investigator used a computer simu- 
lation to test the proposition that "if enough 
monkeys were allowed to pound away at type- 
writers for enough time, all the great works of 
literature would result" (21). The larger objective 
Table 6-4.— Some Examples of Computer Simulation 
of Behavioral Phenomena 
Spacing mechanisms and animal movements: 
• Space use and movement patterns (17,49) 
• Movements of juvenile Atlantic herring (110) 
• Animal spacing (153) 
• Mosquito flight patterns (165) 
• Foraging of the honeyeater bird (169) 
• Random choice in radial arm mazes (194) 
Learning, memory, and problem solving: 
• Classical conditioning (15,18) 
• Learning in neural systems (177) 
• Habituation (195) 
• Behavior in a psychoecological space (84) 
• Mechanisms for reducing inhibition (223) 
Sensation and perception: 
• Visual pattern analysis (13) 
• Landmark learning by bees (37) 
• Chemical recruitment in ants (106) 
Communication: 
• Bird song (55,185) 
• Animal vocalizations (57) 
Sensation and perception: 
• Neuron models (122) 
• Neural basis for pain and touch (148) 
Body maintenance: 
• Food intake (14) 
• Control of drinking behavior (204) 
Reproduction and parental care: 
• Sexual behavior of the male rat (72,205) 
• Infanticide in langurs (89) 
• Evolution of reproductive synchrony (116) 
• Mating behavior of Spodoptera littoralis (200) 
SOURCE: Office of Technology Assessment. 
in this study was to determine if the extreme cases 
of human genius could be accounted for through 
chance processes. 
The simulation was based on an initial assump- 
tion that monkeys typing at random— or a com- 
puter simulation using random numbers— would 
generate huge volumes of nonsense. Statistical 
properties of the English language (e.g., the rela- 
tive frequencies of individual characters or se- 
quences of characters) were added to the simula- 
tion. As higher -order properties of English (i.e., 
the relative frequencies of three- and four-letter 
sequences) were incorporated into the algorithm, 
the rate of generation of intelligible words, phrases, 
and sentences increased . These results led to a hy- 
pothesis that genius could be simulated by a proc- 
ess of random choice with a weighting procedure, 
subject to a prior preparatory process in which 
an individual absorbs the necessary operational 
patterns that characterize the discipline. In studies 
such as this, it is not merely the output generated 
by the computer model that is of interest, but the 
simulation process as well. 
Computer simulations have considerable heuris- 
tic value (65): They may yield insight about the sys- 
tem or phenomenon being modeled (71) and, as 
a consequence, stimulate additional research. The 
value of computer simulation as a heuristic device 
has been summarized as follows (158): 
Simulation gives a means of exploring the plausi- 
bility of models in which theoretical sophistica- 
tion exceeds the state of the art in empirical test- 
ing. Simulations provide tools for empirically 
analyzing theories in order to better understand 
their implications and predictions. Simulations 
are a means of exploring interactions between 
components of complex models. They pose a prac- 
tical challenge to operationalize theoretical con- 
structs, which can lead to incidental discoveries 
about related processes. Finally, they engender 
a concern with issues of process control that con- 
tributes to the development of general principles 
with broad applications. 
Computer simulation holds promise for under- 
standing complex cognitive processes. For exam- 
ple, the computer is often considered analogous, 
at least in some ways, to the human brain (7)— 
both process large amounts of information, and 
their respective outcomes are a consequence of 
