Mountain gorillas use elaborate, multi-stage procedures for dealing with pl
ant defences. This paper investigates the use of mathematically-inspired, i
nformational measures to gauge the complexity of one of these tasks, eating
thistle Carduus nyassanus, from field observations of 38 adults and juveni
les. Behaviour was analysed at two levels, a detailed, movement-based descr
iption of the form of actions, and an organizational description of techniq
ues that were composed of a series of many actions. Complexity, as measured
by counting the sizes of behavioural repertoires, correlated at the two le
vels. Repertoires were shown to be incomplete, but the rates of cumulative
increase in actions differed between tasks. Thistle eating was the most com
plex, and apparently involved many more actions than even chimpanzee tool-u
se. Techniques were highly selective arrangements of actions, so that their
organization (sequence, bimanual coordination, hierarchical structure) ref
lected cognitive capacity. Although ideally it would preferable to estimate
complexity of task organization, this may seldom be feasible, and was not
in this case. Instead, the length of a regularly occurring sequence of acti
ons may be the best practical estimate of an underlying complexity of menta
l process. Confidence in this measure will be increased if it broadly agree
s with other, independent estimates of task complexity; in the case of gori
lla plant processing, both the size of repertoire of functionally distinct
actions and the degree of lateral specialization were, like sequence length
, greater for thistle processing than for other tasks studied to date.