246
The anterior group showed persistence, type I. The lowering of the thres-
hold of activation for recognition units resulted in a degree of arousal to-
wards filter-trills (high frequency of behavioural change) that was not seen
with controls. Thus the anterior group persisted in ‘reacting’ to the test sti-
muli. A test stimulus that in a control bird should produce a slight mismatch,
in the anterior group produced a stronger positive reaction (search and be-
havioural change) as their training-trill specifications had been activated.
Approach was not prolonged for the lower threshold of activation gave rise
to a matching decision similar to that of controls, that resulted in ending the
approach phase.
Persistence, type II, was seen in the posterior group. Damage was more to
the system that would select stimuli likely to match. The birds approached
more stimuli, stayed longer and after the search phase showed more high ten-
sion arousal. They were deficient in the ability to disengage from the situa-
tion. Thus they showed a resulting perseveration of behaviour and generalisa-
tion of response.
In the formulation of this hypothesis the term recognition assumes a match-
ing process between input and representations (specifications) of learned ex-
perience. The concept of matching has a long history of use in models of
orienting and attention systems (Hebb, 1946; Sokolov, 1963; Bernstein, 1969;
Velden, 1978). The term ‘central specifications’ comes from Andrew’s
(1976) reformulation of the term ‘recognition units’ (Treisman, 1960; Kah-
neman, 1973). A recognition unit is one of a population of central mecha-
nisms each of which corresponds to a particular category of stimulus, in the
sense that activation of the unit must occur before responses normally evok-
ed by that category of stimulus are evoked by its presentation (Treisman,
1960; Andrew, 1976). That is to say, when a given activation threshold is
reached, if there is a correct match, the animal will show Consummatory and/
or trained behaviour. If there is a mismatch the animal may disengage or re-
treat from the situation. It is suggested that it is this activation that has been
affected by the anterior group. The threshold for a unit to be activated by
an input has been changed. This threshold is reminiscent of the criterion val-
ue that signal detection theory has suggested must be exceeded for activation
to ensue (Green and Swets, 1966).
The selection of stimuli affects the recognition process. There are rules for
the selection of stimuli. Only a limited number of aspects of a stimulus com-
plex go through to the matching process. A number of characteristics may
have to be used of which no single one is sufficient by itself (cf. response set,
Broadbent, 1970). It is suggested that this selection process has been impair-
ed more in the posterior than in the anterior group as they approached and
searched after more test stimuli than the controls. The impaired selection
occurred at the level of fine call structure and was not a gross sensory deficit,
for the iambus was not mistaken for a trill.