Siamese fighting fish, Betta splendens (Betta),

possess an array of species-specific behaviors (see

Thompson and Sturm, 1965; Bronstein, 1981, 1982,
1985 for review). For example, after opposing males
are driven away, the male Betta establishes a territory
in which he builds a bubble nest which can serve as
a nesting site. Hogan (1961) suggested that females
select mates based on the quality of the males’ territories
and nesting sites. Therefore, by driving away
other males, the chances of reproduction increase for
the defending male.
∗ Corresponding author. Tel.: +1-406-243-20-91;
fax: +1-406-243-63-66.
E-mail address: pyadp@selway.umt.edu (A. Szalda-Petree).
Many researchers have studied the variety of stimuli
that elicit these aggressive behaviors that are characteristic
of male Betta (e.g. Robertson and Sale, 1974;
Thompson and Sturm, 1965; Bronstein, 1981, 1985).
In particular, certain qualities of various neutral stimuli
(colors, shapes, movement, etc.) easily become
associated with unconditioned stimuli that elicit unconditioned
aggressive responses and are preferred in
choice preparations (e.g. Robertson and Sale, 1974;
Bols, 1976).
Robertson and Sale (1974) studied Betta displays
toward eight different models comprised of specific
shapes and body patterns. The Betta displayed aggressive,
submissive, or reproductive behavioral responses
to models styled after male or female Betta.
Subjects displayed the most aggressive behavior in response
to the characteristics of the model styled after
0376-6357/03/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0376-6357(03)00079-2
172 B.B. Craft et al. / Behavioural Processes 63 (2003) 171–175
an aggressive male, specifically featuring long fins,
raised opercula, and no body pattern. Robertson and
Sale noted that a subject’s display toward an aggressive
male model of Betta was indiscriminable from a
display toward a live conspecific and the absence of
any of these features (for example, short fins or a patterned
body) resulted in a different response from the
subject. Therefore, the results of this experiment suggested
that male Betta displayed aggressive behavior
in the presence of certain stimuli (e.g. shape and body
pattern).
Aside from the importance of coloration, shape, and
body pattern, Bols (1976) addressed the importance
of stimulus movements using a submerged T-maze. In
three separate experiments, Bols divided subjects into
two groups, one given the option to choose a Betta
conspecific and another given the option to choose
a Paradise fish (Macropodus opercularis, a related
species). In the first experiment Betta chose between a
conspecific and a nonconspecific (an empty container).
In the second experiment, Betta chose between a conspecific
and a nonconspecific (a marble). Finally, in
the third experiment, Betta chose between a displaying
conspecific and a nondisplaying conspecific. In the
first two experiments, subjects chose the conspecific
more frequently than the nonconspecific. In the final
experiment, subjects chose the displaying conspecific
more frequently than the nondisplaying conspecific.
Bols suggested that conspecifics were more reinforcing
than nonconspecifics due to the conspecific’s display.
In other words, a conspecific performed specific
movements that were lacking in the nonconspecific’s
display.
A review of the aforementioned literature leads to
the deduction that aggressive displays in Betta are
contingent upon specific stimulus qualities. However,
these studies provide no assessment regarding the importance
of stimulus movement. Clearly, Bols (1976)
indicated the importance of stimulus movement in
choice, but Robertson and Sale (1974) stated that a
subject’s display toward an aggressive male model of
a Betta was indiscriminable from a display toward a
live conspecific. Such discrepancies pose a problem in
light of the interchangeable use of mirror images and
live conspecifics in previous research, given that the
two classes of stimuli differ in the specific qualities
of movement. For example, in a mirror presentation,
the movement of the stimulus is completely dependent
on the movement of the subject and may not be
perceived the same as a live conspecific. Currently,
no research has attempted to determine the preference
(if any) between a live conspecific and a mirror
presentation.
If male Betta respond indifferently toward or do
not prefer live conspecifics, then the unconditioned
stimulus fish being used in research could be replaced
with mirrors (halving the number of fish required).
If male Betta prefer live conspecifics, the display or
response of a live conspecific could possibly serve
as a more intense (or less aversive) unconditioned
stimulus presentation. Similarly, if subjects prefer
mirror images, the display or response of a mirror
image could possibly serve as a more intense (or
less aversive) unconditioned stimulus presentation.
The demonstration of a preference could lead to difficulties
with interpretation of previous and future
research.
To explore preference, instrumental conditioning of
choice behavior was used to determine with which
stimulus male Betta prefer to interact. Subjects were
expected to decrease start box latencies and swimway
latencies and choose a live conspecific more often than
a mirror presentation.
2. Method
2.1. Subjects
The subjects (N = 18) were healthy adult male
Siamese fighting fish (B. splendens) obtained from
a local supplier. The naive subjects’ length averaged
6 cm, and the fish were red or blue in color.
2.2. Apparatus
The apparatus was a T-maze similar to the apparatus
used by Bols (1976). The T-maze consisted of
a start/goal box (20 cm × 5 cm × 11 cm), a swimway
(40 cm × 11 cm), and a container for live conspecific
(8 cm×8 cm×11 cm). The T-maze was submerged in
a tank (65 cm × 45 cm × 15 cm; approximately 30 l).
Each tank consisted of a gravel floor, a temperature
gauge, a submerged tank heater, an air stone, and a
T-maze. All latency measures were obtained using a
digital stopwatch.
B.B. Craft et al. / Behavioural Processes 63 (2003) 171–175 173
Fig. 1. Median start box (dashed line) and swimway (solid line) latencies across the 15, 2-trial blocks.
174 B.B. Craft et al. / Behavioural Processes 63 (2003) 171–175
2.3. Procedure
The water used in the apparatus was de-chlorinated
before subjects were introduced and water temperature
was regulated at 25 ◦C throughout the experiment.
Subjects were housed in the entire T-maze during the
experiment and fed a diet consisting of five to six Betta
baby pellets (Hikari, Himeji Japan) daily. Lighting was
also controlled on a 12-h light:12-h dark cycle.
Each subject was tested for two trials each day for
a total of 30 trials. The side for stimulus presentation
was counterbalanced across subjects to eliminate any
potential side bias. Each subject’s start box latency,
swimway latency, and choice were measured for each
trial. Subjects were allowed to remain in the start box
and swimway for a maximum of 10 min before being
forced into the goal box by the researcher. Before the
beginning of each trial, the subject swam or was forced
into the start box and the start box guillotine door was
inserted. Once the choice door, mirror, and live conspecific
were in place, the start box guillotine door was
removed, initiating the trial and the start box latency
measure. Immediately after the subject entered the
swimway, the start box door was replaced, the start box
latency measure ended, and the swimway latency measure
began. Once the subject swam down the swimway
and through the goal box choice door (choosing either
the live conspecific or mirror image), the swimway latency
measure ended. The subject remained in the goal
box for 20 s before the stimulus (the live conspecific
or mirror image) was removed. All doors were then
removed and the subjects were again allowed to swim
freely throughout the T-maze until the next trial.
3. Results
Of the 18 fish, only 16 successfully completed all
30 trials without being forced in the goal box by
the researcher; therefore, all data analyses were conducted
using only 16 subjects. Due to excessive outliers,
Friedman’s Analysis of Variance by ranks was
used to analyze all latency data. Latency data were
collapsed into 15, 2-trial blocks. Friedmans’s ANOVA
revealed a significant decrease in latencies across the
15 blocks for both start box latencies (Fr(k = 15) =
63.87, P < 0.001) and swimway latencies (Fr(k =
15) = 48.69, P < 0.001) (see Fig. 1).
The total number of live conspecific choices was
summed across the 30 trials and analyzed using a
one-sample t-test. Results revealed that subjects made
significantly more live conspecific choices (M =
17.31, S.D. = 3.91) than mirror presentation choices
(M = 12.69, S.D. = 3.91, t(15) = 2.36, P = 0.032).
4. Discussion
In support of the hypothesis, the present study indicates
that subjects’ start box and swimway latencies
decreased significantly over the 30 trials. Thus, the
present study supports previous research that indicates
that an encounter with a live conspecific is reinforcing
(e.g. Bols, 1976; Bols and Hogan, 1979). For example,
in studies by Bols (1976), Betta chose a live
conspecific more frequently than a nonconspecific. In
addition, start box latencies and swimway latencies
decreased over a number of trials when Betta were
given the option to choose a live conspecific.
The present findings revealed a modest, but statistically
significant, preference for the live conspecific
over the mirror presentation. As suggested by Bols
(1976), a possible explanation for subjects’ choice
preference of a live conspecific over a mirror presentation
may result from the conspecifics’ movements. In
a mirror presentation, the movement of the stimulus is
completely contingent on the movement of the subject.
As the subject fluctuates between a submissive display
and an aggressive display in response to the mirror
image, the subject may not encounter a mirror image
with the body pattern, coloration, fin erection, or raised
opercula that elicits aggressive display (e.g. Robertson
and Sale, 1974); thus, the subject may not perceive the
complementary movements of a live conspecific.
The results of the present study indicate a difference
in the Bettas’ perception of a mirror compared
to a live conspecific. The current study, although answering
an important question, was limited to only
examining the preference between a live conspecific
and a mirror presentation; thus, indicating a need for
more research to determine the exact stimuli responsible
for eliciting aggressive behavior and choice in
Betta. Given that previous studies have determined
the effects of different stimuli on conditioned responding
in Betta, further studies could examine the
properties of specific mechanisms (e.g. color, shape,
B.B. Craft et al. / Behavioural Processes 63 (2003) 171–175 175
size, movement, etc.) that elicit aggressive display
and affect the subject’s choice.