BEHAVIORISMroute of working for a living or accepting social
assistance. The arrangement of consequences for
crime and legitimate ways of making a living is
conceptually the same as concurrent schedules of
reinforcement (Hamblin and Crosbie 1977).
Behaviorists are interested in the distribution
or allocation of behavior when a person is faced
with different rates of reinforcement from two (or
more) alternatives. The distribution of behavior is
measured as the relative rate of response to, or
relative time spent on, a specific option. For exam-
ple, a student may go to school twelve days and
skip eight days each month (not counting week-
ends). The relative rate of response to school is the
proportion of the number of days at school to the
total number of days, or 12/20 = 0.60. Expressed
as a percentage, the student allocates 60 percent of
her behavior to school. In the laboratory, a person
may press the left button twelve times and the
right button eight times each minute.
The distribution of reinforcement may also be
expressed as a percentage. In everyday life, it is
difficult to identify and quantify behavioral conse-
quences, but it is easily accomplished in the labora-
tory. If the reinforcement schedule on the left
button produces $30 an hour and the right button
yields $20, 60 percent of the reinforcements are
on the left. There is a fundamental relationship
between relative rate of reinforcement and rela-
tive rate of response. This relationship is called the
matching law. The law states that the distribution of
behavior to two (or more) alternatives matches
(equals) the distribution of reinforcement from
these alternatives (Herrnstein 1961; de Villiers 1977).
Although it is difficult to identify rates of
reinforcement for attending school and skipping,
the matching law does suggest some practical solu-
tions (Epling and Pierce 1988). For instance, par-
ents and the school may be able to arrange positive
consequences when a child goes to school. This
means that the rate of reinforcement for going to
school has increased, and therefore the relative
rate of reinforcement for school has gone up.
According to the matching law, a child will now
distribute more behavior to the school.
Unfortunately, there is another possibility. A
child may receive social reinforcement from friends
for skipping, and as the child begins to spend more
time at school, friends may increase their rate of
reinforcement for cutting classes. Even though the
absolute rate of reinforcement for going to school
has increased, the relative rate of reinforcement
has remained the same or decreased. The overall
effect may be no change in school attendance or
even further decline. In order to deal with the
problem, the matching law implies that interven-
tions must increase reinforcement for attendance
and maintain or reduce reinforcement for skip-
ping, possibly by turning up the cost of this behav-
ior (e.g., withdrawal of privileges).The matching law has been tested with human
and nonhuman subjects under controlled condi-
tions. One interesting study assessed human per-
formance in group discussion sessions. Subjects
were assigned to groups discussing attitudes to-
ward drug abuse. Each group was composed of
three confederates and a subject. Two confeder-
ates acted as listeners and reinforced the subject’s
talk with brief positive words and phrases, provid-
ed on the basis of cue lights. Thus, the rate of
reinforcement by each listener could be varied
depending on the number of signals arranged by
the researchers. A third confederate asked ques-
tions but did not reinforce talking. Results were
analyzed in terms of the relative time subjects
spent talking to the two listeners. Speakers matched
their distribution of conversation to the distribu-
tion of positive comments from the listeners. Ap-
parently, choosing to speak to others is behavior
that is regulated by the matching law (Conger and
Kileen 1974).Researchers have found that exact matching
does not always hold between relative rate of
reinforcement and relative rate of response. A
more general theory of behavioral matching has
been tested in order to account for the departures
from perfect matching. One source of deviation is
called response bias. Bias is a systematic preference
for an alternative, but the preference is not due to
the difference in rate of reinforcement. For exam-
ple, even though two friends provide similar rates
of reinforcement, social characteristics (e.g., status
and equity) may affect the distribution of behavior
(Sunahara and Pierce 1982). Generalized match-
ing theory is able to address many social factors as
sources of bias that affect human choice and pref-
erence (Baum 1974; Pierce and Epling 1983;
Bradshaw and Szabadi 1988).A second source of deviation from matching is
called sensitivity to differences in reinforcement.