LEARNING BY MARINE INVERTEBRATES
33
the sea Panulirus lives in holes among the rocks, and moves mainly at
night. The animals failed to learn to discriminate between the polarized
light sources, which is odd in view of the numerous cases of crustacean
response to polarization plane that are now known (Waterman, 1961).
Schbne’s Panulirus experiment shows that under suitable conditions
decapods rapidly learn by experience. I n the choice of side discrimination an initial error score of more than 70% fell to less than 20% in
twenty trials (in these experiments 50% + errors were usual a t the
start, since an additional error was scored if the animal had to be
prodded into activity on failing to run spontaneously). Performance
continued to improve over a period of weeks.
The experiments of Datta et al. (1960) with the land crab Becarcinus
lateralis Freminville (here considered a marine animal because it goes to
the sea to breed) are less convincing. The authors’ claim to have
“ achieved substantial control over the behaviour of Gecarcinus ”
appears to rest largely on the demonstration that an experimental
group, subjected to a series of reversals in a choice situation, made
substantially more than random errors (see Fig. 20). The control group,
trained in the same situation with both intensity and directional
clues, showed no cumulative improvement in performance over 32
days at ten trials per day. Both controls and experimentals showed
some improvement in performance within each daily run of ten trials
(Fig. 20). Altogether the performance of Becarcinus is unimpressive
compared with that of Schdne’s Panulirus, a rather surprising result in
view of the conditions of the lobster experiments (Panulirus lives
below tide level) and other observations of land crabs in their natural
habitats, which suggest that these animals are by no means slow to
make individual adaptive responses. Datta et al. suggest that their
animals were insufficiently motivated by immersion in freshwater.
B. Other evidence that crustaceans learn
Apart from formal training experiments, there is an increasing
number of observations to show that learning plays an important part
in determining the behaviour of individual crustaceans. Individual
Drornia, for example, appear to have preferred ways of cutting out
coverings for themselves (Dembowska, 1926), locomotion following loss
of parts improves with practice, and crustaceans such as Carcinus and
Homarus avoid traps in which they have once been caught (for references
see Schdne, 1961). Wilson (1949) reports that aquarium specimens of
Palinurus, having once taken to eating Pagurus during a period of
temporary food scarcity, thereafter persisted with attacks on the hermit
crabs which they had hitherto left unmolested.
A . Y . B . 4
0
7
33
the sea Panulirus lives in holes among the rocks, and moves mainly at
night. The animals failed to learn to discriminate between the polarized
light sources, which is odd in view of the numerous cases of crustacean
response to polarization plane that are now known (Waterman, 1961).
Schbne’s Panulirus experiment shows that under suitable conditions
decapods rapidly learn by experience. I n the choice of side discrimination an initial error score of more than 70% fell to less than 20% in
twenty trials (in these experiments 50% + errors were usual a t the
start, since an additional error was scored if the animal had to be
prodded into activity on failing to run spontaneously). Performance
continued to improve over a period of weeks.
The experiments of Datta et al. (1960) with the land crab Becarcinus
lateralis Freminville (here considered a marine animal because it goes to
the sea to breed) are less convincing. The authors’ claim to have
“ achieved substantial control over the behaviour of Gecarcinus ”
appears to rest largely on the demonstration that an experimental
group, subjected to a series of reversals in a choice situation, made
substantially more than random errors (see Fig. 20). The control group,
trained in the same situation with both intensity and directional
clues, showed no cumulative improvement in performance over 32
days at ten trials per day. Both controls and experimentals showed
some improvement in performance within each daily run of ten trials
(Fig. 20). Altogether the performance of Becarcinus is unimpressive
compared with that of Schdne’s Panulirus, a rather surprising result in
view of the conditions of the lobster experiments (Panulirus lives
below tide level) and other observations of land crabs in their natural
habitats, which suggest that these animals are by no means slow to
make individual adaptive responses. Datta et al. suggest that their
animals were insufficiently motivated by immersion in freshwater.
B. Other evidence that crustaceans learn
Apart from formal training experiments, there is an increasing
number of observations to show that learning plays an important part
in determining the behaviour of individual crustaceans. Individual
Drornia, for example, appear to have preferred ways of cutting out
coverings for themselves (Dembowska, 1926), locomotion following loss
of parts improves with practice, and crustaceans such as Carcinus and
Homarus avoid traps in which they have once been caught (for references
see Schdne, 1961). Wilson (1949) reports that aquarium specimens of
Palinurus, having once taken to eating Pagurus during a period of
temporary food scarcity, thereafter persisted with attacks on the hermit
crabs which they had hitherto left unmolested.
A . Y . B . 4
0
7