This study describes a series of visual discrimination experiments in Potamotrygon motoro. Hypotheses included that stingrays would successfully discriminate (a) stimulus-presence from stimulus-absence, (b) different forms, (c) different overall stimulus contrasts, and (d) horizontal from vertical stimulus orientations. All of these hypotheses were supported, although caution must be observed with regard to form discrimination, as will be discussed. The conclusion of an earlier study regarding color discrimination was potentially corroborated, although the colorful stimuli used in the present study were not controlled for with regard to brightness. This report also details the first visual resolution experiment conducted in P. motoro and presents the first evidence of memory retention without reinforcement for this species.
Important limitations of this study are that individual variation between test subjects has a greater influence over results that are obtained from smaller sample sizes and that different trainers conducted each of the visual discrimination experiments described. Both of these conditions were unavoidable, given animal availability and the high investment of time and effort required.
Presence vs. absence
To ensure that stingrays were not blind or cognitively impaired, all experimental animals were tested for the ability to discriminate a black form on a white background from a blank white card prior to further experiments. Only the results for the naïve experimental animals are included in this report, as these are not comparable to the results of experienced animals. All individuals achieved LC, so they were considered fit for further experimentation. Across sessions, the average trial time usually decreased, which likely resulted from animals becoming familiar with the experimental procedure and/or the visual task. There was a potential difference in average trial time across sessions between males and females for the eight stingrays presented with the circle stimulus (Fig. 5), so future studies might consider exploring sex differences in behavioral cognition for P. motoro. Differences in peripheral sensory input have been documented between the sexes in some rays (e.g. Kempster et al. 2013), and sex differences in cognitive behavior have been documented in other fish, such as guppies (e.g. Lucon-Xiccato and Bisazza 2014). However, there is an unfortunate lack of ecological understanding of P. motoro. Based on known differences in parenting roles and sociality in guppies (e.g. Houde 1997; Croft et al. 2004), there is more reason to investigate sex differences in guppies than in these stingrays. Studies to further knowledge of P. motoro ecology would aid in understanding the results of any experiments concerning their cognitive behavior.
Form, color, and contrast
Two different experimental approaches were used to explore whether P. motoro can discriminate form. One group of stingrays (Experiment A) was first trained with two black shapes that were not successfully discriminated and were consequently presented in color. Once the fish learned to discriminate these colorful stimuli, three transfer tests were conducted to determine whether discrimination was based on color/brightness, shape or perhaps both. The second group of stingrays (Experiment B) was trained with two circle stimuli that had different overall contrasts, the design of which was inspired by the species’ dorsal patterning, which is composed of dots with darker, ring-like outlines. Once LC had been achieved, the three stingrays that continued to participate were exposed to two transfer tests, one to confirm that they could discriminate overall contrast and one to investigate whether they had registered information regarding form. Experiment A indicated that stingrays had based discrimination only on color/brightness, while Experiment B indicated that stingrays were able to (a) discriminate contrast and transfer this discrimination to stimuli with unfamiliar form and (b) discriminate a stimulus with the same contrast but a different form from the normally rewarded stimulus. Due to fluctuations in individual performance in the second experiment, it is important to be cautious in interpreting these results, but it would be an obvious mistake at this point to conclude that P. motoro cannot discriminate shapes. Parameters to consider in future studies of form discrimination include what shapes should be used in such experiments. Experiment A used a circle versus cross stimulus pair, and as dissimilar as these shapes may seem to the experimenter, it is possible that the rather radial nature of a cross resembles a circle, dependent on distance and visual acuity perhaps. It seems reasonable to conclude that any attempts to investigate form discrimination, also in other species, should prioritize stimulus simplicity in early experiments. In Experiment B, animals were trained with round forms and then presented, in transfer tests, with angular forms (i.e. triangles), possessing the minimum number of corners for two-dimensional shapes. It should be noted that all forms were symmetrical in this study, so asymmetry was not a confounding factor.
An ability to discriminate stimulus orientation was investigated by training stingrays with horizontal versus vertical stripes and using three transfer tests to elucidate whether discrimination was truly based on stimulus orientation or on overall stimulus images. Stingrays succeeded in the initial discrimination and continued to significantly often choose the horizontal stimuli during transfer tests, indicating that discrimination was indeed based on stimulus orientation and not on overall images. This sort of ability might be useful in identifying vertical or horizontal obstructions or shelters in the riverine environment or perhaps for recognizing the bodily orientations of other animals.
Visual resolution and perception
Riggs (1965) distinguishes between four measures of visual acuity (detection, recognition, localization, and resolution), regarding resolution as most critical. This measure is concerned with an animal’s ability to distinguish the elements of an object or stimulus. To evaluate visual resolution in P. motoro, the present study included a second transfer test phase in which regular trials still presented stingrays with the striped stimuli from the stimulus orientation experiments. Eight transfer tests were conducted in which stripe widths varied from 1 to 10 mm. The narrowest cycle widths that individuals were able to discriminate were used to calculate visual acuity, which ranged from < 0.13 to 0.23 cpd. Behavioral estimates seem to be much lower than anatomical estimates in elasmobranchs (e.g. Ryan et al. 2017), so these low visual acuity values would be expected if the anatomical estimates of 5.52–6.9 cpd for stingrays in the family Dasyatidae (Garza-Gisholt et al. 2015) are shown to also be representative of Potamotrygonidae.
The crescent shape of the P. motoro pupil (Fig. 1) also poses consequences for perception and limits to resolution that should be considered in visual experiments. Murphy and Howland (1991) elaborated on the functional significance of such structures, citing four effects. First, crescent-shaped pupils minimize lenticular spherical aberration by restricting incoming light rays to an equidistant distribution around the lens center; this is in contrast to a circular pupil, which results in a difference between refraction of light rays that pass through the lens periphery and refraction of those passing along the central axis. Second, the presence of the pupillary operculum affects contrast modulation, since an expanding operculum enhances fine details (high spatial frequencies) of a stimulus but reduces the overall stimulus information (low spatial frequencies), including shape. This means the effects of a crescent-shape on contrast modulation and form discrimination may be differentially disadvantageous at certain light levels. Third, Murphy and Howland speculate that the crescent-shaped pupil may function as a focus indicator for organisms lacking a fovea; if the lens focuses in front of an object, points of light reflected off the object appear as ‘U’ shapes, while focusing behind the object results in an inverted ‘U’. Lastly, as a crescent-shaped pupil is constricted, depth of field is reduced while the theoretical limit to resolution is increased, effects which are opposite in a constricting circular pupil. This implies that, at higher light levels, a stingray eye should probably receive images that are more detailed but only over reduced distances. In an attempt to avoid the influence of elevated light levels on results, the present study limited light intensity in the experimental rooms to 320 lx.
In support of the low acuity values found by this study, the environmental and ecological demands on vision for Potamotrygon motoro do not seem to require especially high visual acuity, given the often-elevated sediment loads in riverine waters (e.g. Ríos-Villamizar et al. 2014; Costa et al. 2011), an apparent lack of predators, aside from humans (Charvet-Almeide et al. 2002), and often tactile foraging behaviors (Garrone-Neto and Sazima 2009), though P. motoro’s habitat use and foraging behavior do transition with age (Garrone-Neto and Sazima 2009). The group of P. motoro used in this resolution experiment were already sub-adult, and normally, visual acuity actually improves as fish develop (e.g. Pankhurst et al. 1993). However, larval rainbow trout (Oncorhynchus mykiss) behaviorally demonstrated an acuity of 40 cpd at 10 days after hatching which exponentially decreased to 6.5 cpd at 15 days and 1.4 at 75 days (Carvalho et al. 2004). Adult tuna achieved visual acuity scores < 0.20 cpd (Nakamura 1968), similar to P. motoro.
Studies on larval fish have shown that there is a mismatch between theoretical and behavioral spatial acuity, potentially explainable by the myopia of the larval fish eye (Pankhurst et al. 1993), but, according to Browman et al. 1990, behavioral estimates are probably more accurate than anatomical estimates of visual acuity anyway, since behavior is influenced by a broad range of neurological factors. For example, optokinetic response experiments with larval zebrafish (Danio rerio) resulted in an acuity of 0.16 cpd while the theoretical limit based on anatomy was 0.24 cpd (Haug et al. 2010).
It is unfortunately difficult to compare studies on visual resolution between species, as well as within species, due to differences in experimental approaches, conditions, and animal ages. Various measurements of visual acuity have been derived from optokinetic or optomotor responses, as in the aforementioned rainbow trout studies (Carvalho et al. 2004), or from reaction distances and morphological measurements, as was done in the case of seahorses (e.g. Lee and O’Brien 2011), or from thresholds of stimulus resolution, as with tuna (Nakamura 1968) and the present study. The present study used similar stimuli and the same formula as Nakamura (1968), so those results for tuna are probably the most comparable to our findings for P. motoro. However, Nakamura (1968) used a projector to display the visual stimuli and was therefore additionally able to compare visual resolutions at different stimulus luminance levels between the two tuna species. He found that resolution was comparable at lower luminance but that skipjack tuna (Katsuwonus pelamis) had better resolution abilities at higher values of luminance.
The importance of light conditions regarding a species’ visual apparatus cannot be understated (Murphy and Howland 1991), and neither can the choice of visual stimulus. Honeybees (Srinivasan and Lehrer 1988) and triggerfish (Rhinecanthus aculeatus; Champ et al. 2014), for example, achieved lower scores of visual acuity when presented with radial stimuli instead of linear stimuli. In the previously discussed Form, Color, and Contrast experiments, the present study shows that stingrays seem to pay attention to information about color/brightness over information about shape. Interestingly, Schluessel and Ober (2018) were able to conclude that Potamotrygon motoro prefers directional cues to visual stimulus cards or landmark cues in solving a navigational task. Perhaps it is worth considering that such stimulus preferences could play a role in participation/performance differences between visual discrimination tasks. The experimental design and focus of the present experiment was of course different from that of Schluessel and Ober (2018), but it shows that stingrays participating in stimulus orientation experiments performed more consistently and usually at 100% correct choice as compared to the more fluctuating performance curves of stingrays in other experiments (Fig. 3). Whether or not horizontal vs. vertical stripes can somehow be considered ‘directional’ cues is, however, not clear, and whether differences in performance curves could indicate preferences for certain stimuli is subject to confounding information.
Lastly, it is difficult to determine when an animal has made a stimulus choice. In the present study, each stingray may have made its decision at a different distance from the stimulus wall, but since these distances could not be objectively determined with certainty, a standard distance was used to calculate visual acuity for all individuals. Relatedly, it would be worthwhile for future studies to explore whether the angle of the visual stimulus relative to the stingray has an effect on experimental results. There is no information about photoreceptor or ganglion cell distribution in the eye of P. motoro to date, let alone correlations between the two, but a study comparing closely related stingray species in the family Dasyatidae found that differences in the distribution of retinal neurons seem related to ecology (Garza-Gisholt et al 2015). Future studies with P. motoro should pair behavioral experiments with anatomical investigations and consider exploring whether ontogeny correlates with visual acuity, as in the case of the variable threefin, Forsterygion varium (Pankhurst et al. 1993).
Preliminary tests of memory
An additional pilot investigation into the memory capacity of Potamotrygon motoro was conducted, in which a 14-day break in reinforcement was included during either the stimulus-presence vs. stimulus-absence experiments or the training phase of Form and Contrast Experiment B. Having already achieved LC before the break, two out of four stingrays in the stimulus-presence vs. stimulus-absence experiments were able to quickly (i.e. in less than seven sessions) achieve LC again following the break in reinforcement, indicating that they remembered the task. A fifth stingray was given a break after just two consecutive sessions of ≥ 70% correct choice but officially achieved LC in the first session following the break and maintained ≥ 70% correct choice for two more sessions. This suggests the stingray had already learned the association before LC had been achieved and, furthermore, remembered the association after the break.
Three stingrays in Form and Contrast Experiment B were also exposed to a 14-day break. Having achieved LC prior, one individual achieved LC again after the break, which indicates that memory was not impeded by the use of different, black and white stimuli. The other two stingrays were given a break before they had achieved LC, after only 7 or 14 sessions, to see how soon they would achieve LC thereafter. The individual given a break after 14 sessions achieved LC just four sessions after the break, so the break did not impede the learning process in this individual. The second individual did not achieve LC within 51 sessions, after which the experiment was terminated for that individual. It is, however, highly unlikely that a break in training causes long-term impacts on cognitive ability, so this outcome is probably explainable by individual variation.