A detailed study of the anatomy of the octopus statocyst by Young (1960) seemed to indicate that, although the maintenance of equilibrium was likely to be its most important function, the possibility of its participating in the function of hearing could not be ruled out. The work described here was an attempt to test this possibility.
The general problem of the function of invertebrate statocysts was reviewed by Lowenstein (1950) who pointed out that structural comparisons with the vertebrate labyrinth might be misleading and that direct experimental evidence was needed. Boycott (1960) has described the effects of unilateral and bilateral removal of statocysts on the maintenance of equilibrium in Octopus, and Wells (1960) has described the effects of the same procedure, with its consequent displacement of the retina, on learned discrimination.
The work was done at the Stazione Zoologica in Naples and Octopus vulgaris Lamarck of about 300 g. mass were used. They were placed in pressed asbestos tanks 100 × 60 × 40 cm. deep (Wells & Wells, 1956) through which a continuous supply of aerated sea water was circulated.
The apparatus used for giving sound stimuli through the water consisted of a moving-iron telephone earpiece driven from an audio-oscillator. The earpiece was modified in the following way :
(a) The diaphragm-retaining ring was removed so as to present the maximum vibrational area to the water, and the diaphragm itself was sealed with ‘Araldite’ adhesive to the earpiece all round its outer edge; all screw heads in the surface of the earpiece were also sealed and filled with the same material.
(b) A hexagonal-headed brass bolt, 12 mm. diameter, 5 cm. long, was drilled axially so as to allow a screened twin-core, P.V.C.-covered cable to be threaded through with a tight fit. The outer covering was arranged to lie about 3 mm. back from the face of the hexagon head and the emergent twin wires and the screening were sealed in with ‘Araldite’ adhesive; the exit of the cable at the other end of the bolt was also sealed. The twin wires were connected to the telephone terminals and the screening connected to one of them.
(c) A hollow cylindrical mould whose diameter was about 1 cm. greater than that of the earpiece was made from cold-curing silicone rubber (Midland Silicones Ltd. K9161, Catalyst N9162). This material will allow the release of at least twenty
‘Araldite’ castings. The earpiece and brass bolt were then clamped above the mould and lowered so that the face of the diaphragm rested on three symmetrically arranged small spacers of P.V.C. each about I mm. thick. ‘Araldite D’ to which had been added 8 % by weight of hardening catalyst was then poured into the mould to a level just above the hexagon head and its junction with the shank of the bolt. After about 18 hr. the mould was removed and the unit was ready for fastening to the tank. Fig. 1 shows the various stages of the above procedure. Five earpieces were modified in this way and withstood continuous total immersion in sea water for 3 weeks without breaking down.
The earpieces were mounted singly in the tanks in the middle of the shorter side through a 12 mm. hole drilled at about half the depth of water (about 25 cm.) normally in the tank. A rubber washer and a flat brass washer on each side were sufficient to prevent leakage with only gentle tightening of the fixing nut ; the latter was, however, very firmly tightened to prevent the octopus from turning—and thus loosening—the whole unit (this occurred several times during the installation of the units!).
Fig. 2 shows a block diagram of the electrical apparatus. The audio-oscillator was connected to the earpiece through a shorting key which was opened during the period required for the sound stimulus ; for reasons which will be discussed later no attempt was made to limit the transients in the diaphragm at the beginning and end of the stimulus. Preliminary tests with a calibrated hydrophone in metal tanks rather smaller than those at Naples had shown that many resonances occurred over the range 50-30,000 cyc./sec. ; these were due to the acoustic properties of the earpiece, mismatching at the diaphragm/water interface and reflexions and absorptions in the tank itself. To counteract this, in some measure, each unit was calibrated, before installation, in one of the tanks at Naples, and the oscillator output setting which gave maximum undistorted output over the range 50-30,000 cyc./sec. was recorded. The hydrophone used in these tests was suspended in the tank just in front of the octopus’s ‘home’ (Boycott, 1954) and the output was amplified and displayed on an oscilloscope. Since the absolute calibration of the hydrophone was known it was therefore possible to estimate the effective hydrodynamic sound pressure at any frequency and to see at what level distortion occurred. Boundary effects at the walls of the tank were, however, large, so the figures for stimulus sound pressures were regarded only as giving an order of magnitude rather than an exact absolute figure. Further, this calibration was less useful than might have been expected since there was insufficient time to carry out training tests at more than two or three frequencies and, also, the results obtained were such that the distortion of the wave form from a pure sine curve was not likely to be of much importance. It was found that between 100 and 5000 cyc./sec. the wave-form of the hydrodynamic sound stimulus was undistorted in the neighbourhood of the hydrophone, provided a sound pressure of 17 dynes/cm.2 was not exceeded.
When the earpieces had been installed in the tanks the animals were trained to attack crabs and when the latencies of attack were about 5 sec., one of two further methods of training procedure and testing was adopted.
A crab was placed in the end of the tank opposite the octopus’s home immediately after a short sound stimulus had been given. After five or six repetitions of this at about 10 sec. intervals the sound stimulus was given alone and the animal’s response was noted.
Training to attack with discrimination
First, a 10 V. a.c. electric shock was given to the octopus in its home after a rosee, sound stimulus. Then, after 15 min., the two tests which constituted the training were begun. They consisted of presenting a crab, either alone or preceded by a 10 sec. sound stimulus. In the first case the octopus was allowed to capture— and eat—the crab ; in the second case the crab was withdrawn and, in some experiments, an electric shock was given. The two presentations of the crab were repeated at intervals varying from 5 to 30 min. though in any one experiment the interval between tests was constant. In all cases the measured parameter was the latency of attack by the octopus after the presentation of the crab.
Training to attack
In these experiments three different frequencies—100, 900 and 2000 cyc./sec.— were used both with the same and with different animals on separate occasions; Fig. 3 shows the results from trials with one animal over a period of 9 days. On days 4 and 5 the frequency used was 100 and 2000 cyc./sec., respectively, but on all other occasions it was 900 cyc./sec. The trials were 5 min. apart and the sound was given for either 10 sec.—after which the crab was presented—or for 30 or 60 sec. alone. The 60 sec. period of sound was used on the last 3 days in case responses of more than 30 sec. latency had been missed in the previous tests. Open circles in the figure represent attack latencies when the crab was given; filled circles latencies with the sound alone, 30 sec. being the absolute limit of time measurement.
There was no evidence in these experiments to show that sound conditioning had occurred. It may be seen from the figure that in a total of 130 trials there was only one positive attack out of a possible fifty-seven occasions on which the sound alone was used. There were, however, seventeen occasions (shown as thick vertical bars in the figure) on which the animal showed some movement in its home or emerged slowly from it, usually after the sound was switched off, but none of these movements was comparable with the behaviour of the octopus when it was attacking a crab.
Training to attack with discrimination
Most of these experiments were done with a sound frequency of 100 cyc./sec. and Fig. 4 shows the results of trials on one animal over 5 days. The test intervals were 5 min. and the sound, when used, was given for 10 sec. In the figure the short thick ordinates signify when shocks were given to the animal in its home after a 10 sec. period of sound. No other shocks were given; if an attack was made when the crab was presented after the 10 sec. sound stimulus it was simply withdrawn before it could be eaten. The open circles joined by a continuous line represent attack latencies when the crab was given alone, and the filled circles joined by a broken line represent the latencies when the crab was given after 10 sec. of sound. It may be seen that the graph defined by the filled circles tends to lie above that defined by the open circles; if good discrimination were being performed by the octopus this separation of the two graphs would be more distinct.
Fig. 5 shows histograms of the attack latencies for the two forms of presentation and the inset Table gives the results of t-test calculations; these are given for the whole 5-day period and for the periods between the shocks (See Fig. 4). It may be seen that the mean latency of attack when the crab was preceded by the sound was greater than that without the sound except for one instance ; but there was only one occasion when the difference of these means was significant (P = 0.05). The fact that the distribution is truncated (since the upper limit of measurement was 30 sec.) does not significantly affect the calculations.
It is quite clear that, on the evidence presented here, there is little justification for inferring that the octopus’s behaviour was modified by the presence of sound waves, regardless of whether or not the statocysts were involved. It remains an open question as to whether the results represent the true sensory situation or are a consequence of the particular experimental methods used.
Before a fully satisfactory investigation of any possible sensory faculty of an animal can be made it is important to know what background level of stimulus for that faculty exists in the normal habitat of the animal and, further, what are the physical limits of the naturally experienced stimuli. In this case, therefore, we would like to know what background noise exists in the octopus’s normal environment and what the amplitude and frequency limits of all possible auditory stimuli may be. Since this is at present unknown, the selection of test frequencies (for example) can be no more than arbitrary, although some clues may exist (Dijkgraaf, 1952). It is unlikely that the frequencies used in the present work had much bearing on the negativity of the results, since distortion (and, thus, many harmonics) occurred at sound pressures greater than 17 dynes/cm.2 (see Methods) and this figure was often exceeded; had the results been more positive, pure tones would have been important in estimating the frequency range of hearing.
In this context the definition of ‘hearing’ is of some consequence; Pumphrey (1940) has considered this in relation to insects and his comments are generally applicable although his definition of hearing excluded water-borne sounds. An important point which he made was that the separation of sound and tactile senses was ‘based on the intensity factor’ and was arbitrary. With this in mind there is no reason why we should not define hearing in marine animals as occurring when the animal is ‘demonstratively responsive to sound’ in this case sound being defined as low-intensity hydrodynamic disturbances.
Hence, the hydrodynamic displacements used were about twenty times the corresponding ones in air for human hearing threshold. Pumphrey (1950) has noted the transparency of marine animals to sound (since the densities on either side of the water/animal interface are very much nearer the same value than those between air and terrestrial animals) ; hence the displacements calculated above will reach the statocysts with little attenuation. The phase difference with which airborne sounds reach the mammalian cochlea, however, will be absent at the receptors of marine animals (except at wavelengths comparable with the inter-statocyst distance) and therefore the directional properties of hearing will be lost. Fraser & Purves (1960) have shown how this may be recovered in Cetacea by the acoustic insulation of each ear ; but no comparable structures have been recognized in the octopus.
A special condition which applied to these experiments should be considered here. In any programme which involves the training of Octopus it is essential to maintain good sea-water circulation through the tanks in which the animals are kept (Boycott, 1954). At Naples high oxygenation of the refill liquid entering the tanks was ensured by means of a jet impinging from above on to the water surface. The animal was therefore living in a state of high background noise. This was ‘heard’ qualitatively by the hydrophone but no exact measurement was made of the sound pressures it produced. In order to counteract this, in some measure, the jet was diverted a minute or so before each trial was made (whether or not the sound was used) but this represents a very small fraction of the 24 hr. total circulation. Hence, the octopus may have achieved considerable adaptation to sound which did not necessarily disappear when the jet was diverted.
It may be that a more electrophysiological type of experiment—recording from the statocyst nerve during sound stimulation, for example—would be more rewarding. It would not, of course, answer the question of ‘hearing’ from the behavioural point of view, but it would at least establish whether a mechanism for such a faculty did exist.
A description is given of apparatus suitable for transmitting hydrodynamic sound waves within a rectangular tank.
Two types of training experiment for investigating hearing in Octopus are described, one using simple training to attack, the other training to attack with discrimination.
These results of the experiments, when expressed in latencies of attack on food, gave little evidence that the octopus was able to hear.
Some factors which might have influenced the results are discussed together with suggestions for any future investigations.
I would like to record my appreciation of much technical advice from Mr R. Stubbs and the loan of the hydrophone from the National Institute of Oceanography. Acknowledgement is also due to the Royal Society and the Nuffield Foundation for part of the expenses involved in this work and also to Dr P. Dohrn of the Stazione Zoológica at Naples for laboratory facilities there.