Contrast sensitivity and behavioural evidence for lateral inhibition in octopus

Abstract

Behavioural contrast sensitivity in Octopus tetricus was measured in the range of 0.05–12 cycles per degree (cpd) using a fixation reflex. We show that the contrast sensitivity reaches its maximum (between 1 and 4%) at 0.3 cpd, and decreases to approximately half of the maximum value at the lowest spatial frequency. Reduction of sensitivity at low spatial frequency is a signature of lateral inhibition in visual systems. In vertebrates and insects, lateral inhibition helps to overcome the bottleneck of encoding information into spikes. In octopus, photoreceptors generate spikes themselves and are directly connected to the brain through their axons. Therefore, the neural processing occurring in the octopus brain cannot help overcome the bottleneck of encoding information into spikes. We conclude that, in octopus, either the lateral inhibition occurs in the brain after information has been encoded into spikes, or photoreceptors inhibit each other. This is the first time behavioural contrast sensitivity has been measured in a cephalopod.

1. Introduction

Lateral inhibition is a fundamental feature of visual systems. A classic example of lateral inhibition is the processing of visual information in vertebrate retina and insect lamina by cells having centre–surround receptive fields [1,2]. Barlow [3] suggested that lateral inhibition can be explained using information theory. In order to fit information into a channel with limited capacity, the signals need to be encoded in an economic way, namely, redundancy from signals needs to be removed [4–7]. In vertebrates and insects, lateral inhibition occurs at the first synapse of photoreceptors before spikes are generated and therefore minimizes the stream of synaptic vesicles from photoreceptors [2,8,9]. Because conversion from graded potential to spikes generally reduces information capacity [10,11], spike generation is likely to form a bottleneck for information transfer. In vertebrates, information is sent from retina to brain as a message of spike trains via the optic nerve, which has limited information capacity [4,5].

Similarly to vertebrates, cephalopods have a camera-type eye, which is connected to the brain via the optic nerve. However, the cephalopod eye does not have a neural retina. Instead, cephalopod photoreceptors generate spikes themselves and their axons form the optic nerve [12,13]. The area of the cephalopod brain where the optic nerve projects—the plexiform layer of the optic lobe—is thought to be analogous to the vertebrate neural retina, and is called ‘deep retina’ [14,15]. Unlike the vertebrate retina, the cephalopod plexiform layer does not encode photoreceptor signals into a compact message of spike trains. Therefore, the neural processing in the cephalopod plexiform layer cannot be explained by the need to overcome the bottleneck of optic nerve and spike generation, but it still may decrease the flow of synaptic vesicles from neurons in the octopus brain.

A behavioural manifestation of lateral inhibition in visual processing is a reduction of contrast sensitivity to low spatial frequencies owing to the opposing signals of centre and surround of receptive fields [16]. Indeed, vertebrate and insect behavioural contrast sensitivity decreases at low frequency [17–21]. However, behavioural contrast sensitivity has not been measured in cephalopods [22]. Here, we report contrast sensitivity in a cephalopod (Octopus tetricus, Gould, 1852).

In octopus, the reduction of sensitivity at low frequency may indicate either that lateral inhibition occurs in the brain after the bottleneck of spike generation or that inhibitory interactions occur between photoreceptor axons owing to interconnecting collateral nerve fibres. This has been previously described in Limulus sp. [23], but in cephalopods such interactions have been hypothesized but not demonstrated [12–15,22,24]. In contrast to Limulus, octopus collateral fibres do not form a dense mesh, and the cell bodies of some collateral fibres reside in the optic lobe [13–15].

2. Methods

Five individuals of Octopus tetricus were used: two for preliminary experiments and three for measuring contrast sensitivity. Stimulus detection was determined by a ‘fixation response'—expansion of the chromatophores in and around the eye during accommodation, resulting in a momentary darkening around the pupil—[25] (figure 1b; electronic supplementary material, movie S1). The fixation occurs when the octopus attends to a visual stimulus, and does not require training. This fixation reflex is the ‘dark eye ring' behaviour seen in octopuses when disturbed, which is commonly expressed when an object in the water suddenly appears or approaches the octopus [26]. After about four to five weeks, octopuses stopped attending to the stimuli—they hid behind rocks or covered their eyes with shells. This limited the duration of experiments with an individual animal.

Figure 1.

(a) Diagram of the experimental set-up. The black corrugated plastic screen limits the field of vision to the LCD monitor and impedes any other visual stimulus outside the tank from reaching the octopus's eyes (figure not to scale). (b) Octopus fixation response. (i) Octopus eye when stimulus is not present. (ii) Octopus eye when stimulus is present. The pupil expands along with chromatophores in and around the eye, resulting in a slight darkening around the pupil.

Vertically oriented sinusoidal gratings (12, 4, 1, 0.3, 0.1 and 0.05 cycles per degree (cpd), temporal frequency of 2 Hz) were presented over a 50% grey background (0.101 cd m⁻²) on a 19 inch LCD monitor emitting 45° linearly polarized light and running at 75 Hz (DELL UltraSharp 1907FP), at a distance of 1 m from the tank (figure 1a). Stimuli were presented for 2 s with a 20 s interval between presentations. A positive response was recorded if a fixation reflex was detected no later than 3.5 s after the onset of the stimulus. Preliminary experiments showed that all frequencies were detectable at 100% contrast. Frequencies were presented in random order. False alarm rate was determined as the number of fixation reflexes outside of the interval of stimulus presentation (3.5 s) during a 10 min trial where stimuli were presented at 100% contrast, and the proportion of blind responses was determined as the lack of response to a stimulus with 100% contrast. A false alarm rate of 0.057 and a proportion of blind responses of 0.306 were found (see electronic supplementary material). It is important to note that false alarms could be present due to a variety of circumstances, such as the octopus focusing on debris floating in the water or simply accommodating to scan the environment at different depths, or due to neural noise.

Contrast thresholds were recorded using an adaptive staircase with a Weibull approximation of the psychometric function (QUEST [27]) and contrast sensitivity was defined as the inverse of the threshold. To avoid attenuation resulting from high contrast stimuli, the staircase procedure had an initial contrast of 5% and was capped at 10%. Therefore, the values of sensitivity below 10 could not be estimated reliably. Minimum contrast, based on hardware limitations, was 0.62%. Contrast sensitivity was fitted by a difference of Gaussian and exponential functions [28], a model that accurately describes contrast sensitivity of humans [29]. Error bars for contrast sensitivity and 25 and 75% quartiles (Q_1–3) for the receptive field sizes were calculated using a bootstrap procedure [30]. It is important to note that we recorded the threshold for attracting attention rather than the minimal detectable contrast, as has been done in studies where human observers maintained attention throughout the experiment [29].

3. Results and discussion

In all three animals, the sensitivity decreased at low frequency and peaked at 0.3 cpd (figure 2). By comparison, human photopic contrast sensitivity peaks at 3 cpd [29]. Previous studies demonstrated that octopuses can be trained to discriminate high contrast (black and white) gratings from grey [31–33]. Sutherland [33] found that the maximum spatial frequency that Octopus vulgaris can detect is 1.76 cpd, while Muntz & Gwyther [31,32] found that that maximum spatial frequency, for Octopus australis, is in the range of 3–6.8 cpd and, for Octopus pallidus, is 4.23 cpd. Our preliminary experiments demonstrated that octopuses can detect up to 12 cpd, which is significantly higher than previously reported. Anatomical estimates of the maximal frequency that can be resolved without aliasing (Nyquist frequency) based on a distance between photoreceptors of 6.5 µm and an eye diameter of 1.3 cm [34] give 12.5 cpd (see electronic supplementary material, Nyquist frequency calculation), which is very close to the maximum frequency that octopuses could detect in our experiments, while Land gives a significantly higher estimate based on a photoreceptor distance of 3.5 µm [35]. Note that the Nyquist frequency for human fovea is 56 cpd [36].

Figure 2.

Contrast sensitivity functions for three octopuses. Error bars indicate 25 and 75% quartiles. The line represents the best fit of the centre–surround model.

The bell-like shape of contrast sensitivity can be explained by a model based on properties of vertebrate ganglion cells with centre–surround receptive fields [28]. At the highest frequency (12 cpd) the estimated sensitivity was less than 10—the lowest sensitivity that could be estimated reliably. Therefore, this frequency was omitted for model fitting. The model gives the following estimates of the half-widths of the centres: 0.3° [Q_1–3 0.13–0.34], 0.09° [Q_1–3 0.085–0.095], 0.06° [Q_1–3 0.05–0.07] and surrounds: 1.8° [Q_1–3 1.3–4.0], 2.1° [Q_1–3 1.8–2.6], 1.9° [Q_1–3 1.7–2.3] for octopuses 1, 2 and 3, respectively. The maximum contrast sensitivity corresponded to thresholds of 0.97, 3.38 and 2.61% for octopuses 1, 2 and 3, respectively. This variation may be due to the differences in attention thresholds and to the differences in false alarm rates and blind response proportions, which have been estimated in preliminary experiments with different animals. The octopus thresholds are comparable to the minimal detectable contrast of human observers (approximately 1%) [29]. However, considering we recorded the threshold for attracting attention rather than the minimal detectable contrast, cephalopods possibly can detect contrasts that we are unable to see. Previously it has been demonstrated that cuttlefish displays disruptive body pattern as a response to a chequerboard starting from 15% contrast [37], which corresponds to a much higher threshold than we have revealed in octopus using a fixation response.

The reduction of sensitivity at low frequency provides strong evidence in favour of the hypothesis of lateral inhibition occurring either due to neural processing in the octopus brain, after information has been encoded into spikes, or due to the interaction between photoreceptors via collateral fibres [12–15,22,24], as has been demonstrated in Limulus [23].

Ethics

All animal handling and experiments were carried out under the approval of the University of Auckland Animal Ethics Committee (ref. 001761).

Data accessibility

Data are uploaded as electronic supplementary material and to the Dryad Digital Repository (doi:10.5061/dryad.2c984j0) [38].

Authors' contributions

L.N.-R. performed the experiments. M.V. conceived the study and modelled the contrast sensitivity curves. Both authors analysed and interpreted the data and wrote the manuscript. Both authors contributed to revising the manuscript and approved the final version of the manuscript. Both authors agree to be held accountable for the content of this work.

Competing interests

The authors have no competing interests.

Funding

This work was funded by Consejo Nacional de Ciencia y Tecnologia, Fellow CVU no. 516833.