233
rnal training trill (122 cps) filtered by 40 dB under 400 Hz (Tj, 55 dB), 3.15
KHz (T3, 50 dB) or 8 KHz (T5, 50 dB), played at slow speed (T2, 63 cps, 60
dB) or fast speed (T4, 160 cps, 66 dB), an ‘arousal’ trill (T6, 127 cps, 64 dB),
the winding of a watch (T7, 29—53 cps, 61 dB), a novel food-trill (T8, 127
cps, 54 dB) and an iambus-song call (T9, 59 dB). The number in parentheses
refers to the order of presentation and the average sound pressure level in the
test chamber is given. No response to a test call was rewarded. The test pro-
gramme consisted of nine test stimuli. Between each test stimulus two train-
trills were played so that no effect of one test stimulus would effectively car-
ry over to the next test stimulus. Time intervals between stimuli were 1, 2, 3
or 4 min in a pseudo-random order.
The behaviours recorded were feeding, move (move feet — M), peck floor,
drinking, shake∕feather ruffle, presence by dish, search (S), low tension (LT)
and high tension (HT) postures. Search behaviour was characterised by the
extension of the neck with the head still and/or with side to side movements
of the neck below the body line (Fig. 3c). In high tension postures the legs
were straighter and the neck was extended vertically. The head would some-
times turn but without sideways movement of the neck (Fig. За). In low ten-
sion postures the neck and legs were relaxed (Fig. 3b).
Analysis
Preliminary analysis showed that the data were not normally distributed.
Variance of preoperative populations was considered (e.g. approach duration,
0.95 > P> 0.9, Kruskal Wallis analysis of variance). Thus non-parametric tests
appropriate for the behaviour of small samples were used (Ferguson, 1966;
Siegel, 1956). Pre- to postoperative changes in the overt behaviour rather than
absolute values were analysed. The data of the experimental groups were
compared with preoperative and prestimulus conditions (related sample tests)
and with sham-operates (unrelated sample tests). The probabilities given are
for the performance between groups where Kruskal Wallis analysis of variance
showed there to be changes; the Mann Whitney U test confirmed the groups
responsible for the change (Siegel, 1956). Preoperative observation showed
that overt behaviour occurred in phases (see Overt behaviour, p. 235). Thus
a rank analysis for dominant behaviour in a phase was computed.
RESULTS AND DISCUSSION
Control measures of the incidence of the four main overt behaviours (M,
S, LT, HT) were taken with undisturbed birds in the presence of a novel ob-
ject (brown bottle) and birds with the novel object and food for 5 min under
each condition. There were no significant differences between groups.
The pattern and sequence of behaviour associated with the train-trill will
be described in the first section. The analysis concerns the 60 s after a stimu-
lus. The occurrence of the behavioural components seldom reached prestimu-
lus levels sooner.