Vision in two cyprinid fish: implications for collective behavior

Visual field configuration

A variety of different approaches, both direct and indirect have been used to measure the visual field configuration of fish (Watanuki et al., 2000; Bianco, Kampff & Engert, 2011). However, direct estimates are often more precise, as they take into account the spatial extent of the retina and its projection into the visual space (Martin, 1984; Martin, 2007). For large species with large eyes that can tolerate supplemental ventilation and the placement of electrodes into the eye, the visual field can be measured with an electroretinogram on live animals (McComb & Kajiura, 2008; McComb, Tricas & Kajiura, 2009; Kajiura, 2010). However, for smaller species with small eyes, like the ones used in this study, ophthalmoscopic measurements are more appropriate (Collin & Shand, 2003). We measured the visual field configuration of 5 adult zebrafish and 9 adult golden shiners using an ophthalmoscopic reflex technique (Martin, 1984) and modified visual field apparatus developed for fish (Appendix S1). This technique requires the direct observation of the eye with an ophthalmoscope (Keeler Professional; Keeler Ophtalmic Instruments, Broomall, Pennsylvania, USA) to detect the light reflected from the retina, so observations were not made with the animal in water (Appendix S1). Individuals were measured immediately following euthanasia via rapid cooling through submersion in a 5 parts ice/1 part water mixture (Wilson, Bunte & Carty, 2009). The apparatus was configured such that 90° and 270° were directly in front and behind of the animal, respectively, along the horizontal plane, while 0° and 180° represented the vertical plane above and below the animal (Appendix S1). We recorded the visual field configuration across all angles around the head in 10° elevation increments. Our estimate of the visual field did not take into consideration variations due to eye movement, but considered the resting eye position, as previous studies suggest that the eyes assume a resting position postmortem (Martin, 1986). Additionally, the resting eye position is maintained in between convergent and divergent eye movements, and is thought to remain stabilized by the vestibulo-ocular reflex as the animal swims (Easter & Johns, 1974; Schairer & Bennett, 1986).

Based on the combined measurements of both the left and right eye at each elevation, we estimated the species’ binocular, lateral and blind area sectors (Martin & Katzir, 1999). The binocular field was determined separately at each elevation as the overlap in vision between both eyes while the blind area represented regions devoid of vision. For the lateral area, we applied the following equation: (360—(mean blind area + mean binocular field)/2) (Fernández-Juricic et al., 2008). Averages for each species were determined at each elevation with accuracy of ±0.5°.

Retinal wholemounting

We extracted and wholemounted the retinas of 3 adult zebrafish and 3 adult golden shiners. Following euthanasia via rapid cooling (Wilson, Bunte & Carty, 2009), we submerged the head of the animal in Davidson’s Fixative for 3–5 h. While the eye remained attached to head by the optic nerve, we used spring scissors to remove the cornea, lens and vitreous humor. A dorsal cut was made to the retina before the eyecup was detached to ensure the correct orientation in later steps. We separated the retina from the eyecup by peeling away the sclera with forceps. The retina was then subjected to a (3%) bleaching solution until transparent to allow for the visualization of the ganglion cells. Following bleaching, 3–4 peripheral cuts were made to allow the retina to lie flat on a gelatinized glass slide. We followed the procedures described in Ullmann et al. (2012) to fix and stain the retina using a series of dehydrating and rehydrating steps in combination with 0.1% cresyl violet. In addition, we photographed images of the wholemounted retina and recorded area and perimeter measurements to account for retina shrinkage before and after staining using ImageJ (http://imagej.nih.gov/ij/). Average retina shrinkage values for the zebrafish and golden shiner were 0.07 ± 0.05 and 0.13 ± 0.07 respectively (mean ± SE).

Following staining, the retinas were viewed under an Olympus BX51 microscope. We traced the perimeter of the retinal periphery and optic nerve head with Stereo Investigator v.10 (MBF Bioscience, Williston, Vermont, USA) and then applied a sampling grid with a counting frame of 50 µm² × 50 µm² for a total of approximately 200 total sampling sites per retina. We then counted the total number of observed retinal ganglion cells at each of the sampling sites using ImageJ (http://imagej.nih.gov/ij/). Ganglion cells were distinguished from amacrine and glial cells according to previously established morphological characteristics (Hughes, 1977; Ehrlich, 1981; Collin & Pettigrew, 1988). We omitted counting sites where the retina appeared damaged or the cell visibility was poor. For the zebrafish, we counted 150 ± 54.9 sites per retina; while for the golden shiner we counted 112 ± 15.7 sites (mean ± SE). The Schaeffer coefficient of error for the zebrafish and golden shiner was 0.006 ± 0.0015 and 0.003 ± 0.0002 respectively (Glaser & Wilson, 1998). Topographic maps were generating using the R program “one cell map V8 svg version.R” (Garza-Gisholt et al., 2014). This program constructs isodensity maps from regions of localized cell density within a defined retinal outline created with Adobe Illustrator.

Spatial resolving power

Retinal ganglion cells act as a bottleneck filter, summating the information from the other retinal cell types (i.e., bipolar cells, amacrine cells, photoreceptors) into a signal that is sent to the brain (Pettigrew et al., 1988). Considering the integral role that the ganglion cells play in relying visual information to higher processing centers, the anatomical estimates of spatial resolving power based on ganglion cell densities and eye size have been used extensively in the literature (Hughes, 1977; Collin & Pettigrew, 1989; Collin, 1999; Pettigrew & Manger, 2008). In addition, this anatomically derived measure of spatial resolving power has been shown to closely resemble behavioral estimates of visual acuity in species of fish, mammals and birds (Hughes, 1975; Hughes, 1977; Pettigrew et al., 1988; Pettigrew & Manger, 2008; Temple, Manietta & Collin, 2013).

In fish, the lens provides the primary source of focusing power due to the large dissimilarity in the refractive index between the aquatic environment and the optical media (Bone & Moore, 2008). The lens diameter also provides an index of eye size and body length, as it has been shown to increase in diameter as the fish matures (Pankhurst, Pankhurst & Montgomery, 1993). Consequently, calculations of spatial resolving power in fish incorporate lens dimensions (Collin & Pettigrew, 1989) as opposed to similar estimates in other vertebrates where eye axial length is generally used (Hughes, 1977; Martin, 1986). We followed (Collin & Pettigrew, 1988; Collin & Pettigrew, 1989) in the approach to estimate spatial resolving power using retinal ganglion cell density. For teleosts, the posterior nodal distance (PND) is calculated by multiplying 2.55 by the radius of the lens to estimate the area that an object would take up on the retina. Using the calculated PND, we then followed the equation from Pettigrew et al. (1988) and Williams & Coletta (1987) to estimate the retinal magnification factor, $RMF=\frac{2\pi *PND}{360}$ and anatomically derived spatial resolving power (SRP), where D is the density of retinal ganglion cells (cells/mm²): ${\displaystyle SRP=\frac{RMF}{2}\sqrt{\frac{2D}{\sqrt{3}}}.}$ Using the estimated spatial resolving power of these cyprinid species, we calculate the extent of their ability to resolve objects in the environment. We followed the equation from Tyrrell et al. (2013) to estimate the threshold distance (d) at which an object exhibiting maximum visual contrast can be resolved under optimal light conditions, where r is the radius of the object and α is the inverse of spatial resolving power: ${\displaystyle d=\frac{r}{tan\frac{\alpha }{2}}.}$ With this equation, it is possible to estimate the threshold maximum distance that an animal can resolve an object across various regions of the visual field. However, these distances do not incorporate the sensitivity and distribution of the photoreceptors, which also are also responsible for contrast sensitivity under different light environments (Lythgoe & Partridge, 1989). For the purposes of this study, we used the peak spatial resolving power (i.e., area with maximum ganglion cell density) and the minimum spatial resolving power (i.e., area with minimum retinal ganglion cell density) to calculate the maximum distance at which the body of conspecific could be resolved for both species in different regions of the visual field. In addition, we also estimated the maximum resolvable distance for the species-specific conspicuous markings, as it has been suggested that social fish use characteristics of the conspecific phenotype, such as spots or stripes as reference marks to assess changes in neighbor movements when schooling (Guthrie & Muntz, 1993; Bone & Moore, 2008). Wildtype zebrafish possess alternating light and dark horizontal stripes along the sides of their bodies, which may be used as a cue to assess the separable distances of group mates when shoaling (Rosenthal & Ryan, 2005). Golden shiners contrastingly do not have conspicuous stripes like the zebrafish; however, they do possess defined cycloid scales, which have smooth edges that could allow animals to distinguish individual scales. Therefore, we calculated the maximum resolvable distance for the golden shiner scales, as in some species the reflective scales act as a visual cue during social interactions (Rowe & Denton, 1997). We measured photographs of the zebrafish wildtype striping pattern (four individuals) digitally using ImageJ (http://imagej.nih.gov/ij/) and estimated the length of one cycle to be the height of two alternating, horizontal stripes, one dark and one light, assuming that both stripes were equal in dimension. Using similar procedures, we measured the average visible width of the cycloid scales of four golden shiners.

Position and projection of the center of acute vision into the environment

We followed the protocol of Moore et al. (2012) to estimate the location of the center of acute vision on the flattened retina using Cartesian coordinates: -X, Y (dorsal nasal), X, Y (dorsal temporal), -X, -Y (ventral nasal), and X, -Y (ventral temporal). Coordinates were averaged across 3 individuals per species to determine the location of the center of acute vision along the four retinal axes (nasal, temporal, ventral and dorsal).

Additionally, using the monocular visual field of each eye and the ganglion cell topographic maps, we estimated the angular projection of the center of acute vision into the environment. Firstly, this required restoring the retina to its original shape because in the process of wholemounting we made several cuts to flatten it. The peripheral cuts of the topographic map were stitched together digitally using the R package, “Retistruct” (version 0.5.7) (Sterratt et al., 2013). The program uses a mathematical algorithm to digitally fit an object onto a spherical globe. The user provides the object of interest (i.e., retinal topographic map) and specifies the orientation in addition to other defining features of the retina, such as the optic disc (i.e., optic nerve head). We specified in the program the orientation of the dorsal cut applied to the retina during extraction. Following the reconstruction, we determined the angular position of the projection of the center of acute vision into the environment by reflecting the polar coordinates generated by the program into corresponding locations of each species’ visual field. These coordinates utilize a colatitude and longitude system where 0° lies at the center. The nasal and dorsal poles are −90° and the temporal and ventral poles are +90°.

Visual field configuration

In the horizontal plane of the head (90°–270°), the binocular field width of the zebrafish and golden shiner were 33 ± 3.54° and 31 ± 1.67°, respectively (Figs. 1A and 1B). The height of the binocular field extended well above and below the horizontal plane of the head in both species (Figs. 1C and 1D). In the zebrafish, the upper portion of the binocular field projected to a point right above the head, 0° (Fig. 1C). However, in the golden shiner, the height of the binocular field extended an additional 40° behind the head (Fig. 1D), providing wider visual coverage in the vertical plane.

In the horizontal plane of the head, the zebrafish and golden shiner had a blind area of 21 ± 2.30° and 25 ± 1.20°, respectively. The size of the blind area limited the visual coverage (binocular plus lateral fields) in the horizontal plane to 339 ± 4.0° in the zebrafish and 335 ± 3.5° in the golden shiner. Considering all recorded elevations, the average visual coverage of these cyprinid species was: 336 ± 3.8° for the zebrafish, and 333 ± 3.5° for the golden shiner.

Retinal ganglion cell density and centers of acute vision

The density of retinal ganglion cells was not homogeneous across the retina (Fig. 2). Both species had a high density of retinal ganglion cells in the ventro-temporal region of their retinae and the pattern of cell density increase was concentric (Fig. 2). The characteristics of these areas of high retinal ganglion cell density (i.e., lack of morphological features associated with retinal invagination on the wholemounted retina, concentric increase in cell density) suggest that the centers of acute vision of both species are areae (i.e., enlargement of the retinal tissue due to the high density of photoreceptors and retinal ganglion cells; Walls, 1942). The zebrafish had a peak ganglion cell density of 36,224 ± 730 cells/mm², while the golden shiner a peak density of 14,380 ± 1272 cells/mm². In order to quantify the change in cell density across the retina, we utilized a concentric sampling approach to best reflect the characteristics of the area (Figs. 3B and 3C). The cell density difference between the center of acute vision (i.e., maximum density) and the retinal periphery (i.e., minimum density) was 18.29 ± 2.41% in the zebrafish and 17.55 ± 2.86% in the golden shiner on average (Fig. 3A).

Based on the ventro-temporal location of the centers of acute vision (Figs. 2A and 2B), we can assume that both species possess high acuity vision in the fronto-dorsal region of their visual field (Figs. 2C and 2D). For the zebrafish, the center of the area projected to −32 ± 15°, −64 ± 10° (nasal, dorsal) of the visual field while the center of the area projected to −20 ± 8°, −9 ± 4° (nasal, dorsal) of the visual field in the shiner. However, the location of centers of acute vision on the retina was slightly different in both species (Figs. 2C and 2D). For the zebrafish, the center of the area was more ventrally shifted and located −0.10 ± 0.06 along the nasal-temporal axis and −0.21 ± 0.11 along the dorso-ventral axis. In the golden shiner, the center of the area was more temporally shifted and located −0.01 ± 0.12 along the nasal-temporal axis and −0.17 ± 0.04 along the dorso-ventral axis. Consequently, the fronto-dorsal projection of the area varied between species.

Spatial resolving power and maximum resolvable distance

In terms of eye measurements, we recorded the average lens diameter (mm) of both species as 0.83 ± 0.03 for the zebrafish and 2.00 ± 00 for the golden shiner; the eye transverse diameter was 1.91 ± 0.04 and 4.53 ± 0.20; and the eye axial length was 1.35 ± 0.10 and 3.20 ± 0.23, respectively. Additionally, the average total body length (cm) was 4.17 ± 0.07 for the zebrafish, and 7.4 ± 0.06 for the golden shiner.

According to the peak density of retinal ganglion cells and lens size, the visual acuity of the zebrafish and golden shiner were estimated to be 1.89 ± 0.02 and 2.86 ± 0.12 (cycles/degree), respectively. With their highest acuity, provided by the center of acute vision, zebrafish could resolve the body a conspecific up to 75.80 cm (18 body lengths) away and the striping pattern up to 19.49 cm (5 body lengths) away, while golden shiners could resolve a conspecific up to 2.92 m (39 body lengths) away and the cycloid scales of a conspecific up to 19.66 cm (3 body lengths) away. In the peripheral region of the retina with the lowest cell density, visual acuity was estimated as 0.08 ± 0.06 cycles/degree for the zebrafish, and 1.18 ± 0.15 cycles/degree for the golden shiner. Based on this retinal periphery estimates, a zebrafish could resolve a conspecific and the striping pattern up to 32.08 cm (8 body lengths) and 8.25 cm (2 body lengths), respectively; while the golden shiner could resolve a conspecific and the cycloid scales up to 1.20 m (16 body lengths) and 8.11 cm (1 body length), with the periphery of the retina.

Implications for collective behavior

Based on the characteristics of the visual field configuration and retinal topography reported in this study, we developed species-specific predictions relative to their collective behavior (i.e., positioning, orientation and spacing of individuals in a group) that can be tested empirically in the future.

Both species have a wide range of visual coverage limited by a posterior blind area, which may make them vulnerable to predator attacks. However, individuals may be able to increase their “collective” visual coverage by associating in groups where the vigilance can be shared among conspecifics (Rountree & Sedberry, 2009). In other words, an individual’s blind area may be compensated for by the visual coverage of its surrounding neighbors, which have the capacity to provide cues to the individual about potential dangers. However, this degree of visual compensation would vary depending on the spatial arrangement of the group members. For example, in a school, where all of the members are facing the same direction, the individual blind areas are spatially located behind the school (Fig. 4). Alternatively, in shoals, group members are not necessarily facing the same direction; therefore, the individual blind areas may project to different regions around the group. In some instances, a “collective” blind area (e.g., red area in Fig. 4) is generated in which all of the individual blind areas overlap, and represents a blind zone that is shared between all of members of the group.

The size and location of the “collective” blind area could have implications for the group’s ability to detect approaching predators (Rountree & Sedberry, 2009). As individuals space themselves out, the distance between the center of the school and the “collective” blind area increases, providing individuals with more time to detect and respond to potential predators approaching from behind (Fig. 4), assuming equal predator attack speeds. Alternatively, as individuals become closer in the group, the distance to the “collective” blind area decreases, reducing the amount of time to respond to a predator attacking from behind (Fig. 4).

Using the blind areas of both cyprinid species reported in this study, the average neighbor separation (Burgess & Shaw, 1981; Miller & Gerlai, 2012) and group sizes observed in the wild (Werner, Hall & Werner, 1978; Pritchard et al., 2001), we calculated the horizontal distance from the center of the school to the “collective” blind area following the equation presented in Rountree & Sedberry (2009). In addition, we used these calculated distance ranges to estimate the time that both species would have available to react to a predator originating from the starting point of the “collective” blind zone (Fig. 4). These estimates assume all individuals are the same size, are oriented in the same direction, and maintain the same nearest neighbor distance.

In the wild, zebrafish are known to form shoals consisting of 2–10 individuals (Pritchard et al., 2001) with nearest neighbor distances ranging from 0.5 to 5.5 cm (Miller & Gerlai, 2012). Based upon these behaviorally measured nearest neighbor distances, we would expect the distance to the “collective” blind area to range from 0.01 to 1.34 m behind the school (Figs. 4A and 4B). Golden shiners have been observed in schools up to 45 individuals with nearest neighbor distances ranging from 9 to 20 cm. Considering this neighbor separation distance, the distance to the “collective” blind area would range from 8.93 to 19.85 m behind the center of a golden shiner school (Figs. 4C and 4D). Considering these relationships, individuals may tradeoff having long interaction distances to increase predator detection with short interaction distances, to allow for the improved resolution of conspecific cues. In fact studies, on zebrafish seem to support this idea as, groups often decrease their neighbor separation immediately following predator exposure (Miller & Gerlai, 2007).

The freshwater garfish (Xenentodon cancila) is one of the most prevalent predators of zebrafish (Spence et al., 2008) and are known to rapidly pursue prey (Hossain et al., 2013). We calculated that the zebrafish could potentially increase the time available to react by 13,300% by widening the size of the school from of 0.01 m to its maximum 0.50 m by increasing in neighbor distance from 0.5 to 5.5 cm (0 to 1 body length), which is the range assumed by species in the wild and is also within the range to perceive conspecific cues (Miller & Gerlai, 2012). Largemouth bass (Micropterus salmoides) are common predators of golden shiners (Reid, Fox & Whillans, 1999) and by increasing the school width from 3.96 m to 8.80 m with an increase in neighbor distance from 9 to 20 cm (1 to 3 body lengths), golden shiners could potentially increase their reaction time to this predator by 122%. Similarly the neighbor distances used in this calculation represents a range that would allow individuals to perceive conspicuous markings and has also been behaviorally measured in golden shiners (Burgess & Shaw, 1981). In active pursuit predators, like the billfish, which target large schools of prey, increasing the visual range behind the school may improve the survival of the individual group members, as it may allow members to detect rapid changes in the predator’s movements (Hobson, 1979; Pitcher & Parrish, 1993). However, the predator detection benefits of increasing neighbor distance should be outweighed by the costs of reducing the benefits of dilution effects.

In both species, the centers of acute vision are directed towards the front-dorsal region of their visual field (Figs. 2C and 2D). When traveling in a school, zebrafish and golden shiners may prefer positions where they can align this high acuity region with conspecifics to enhance group cohesion. Zebrafish may preferentially respond to neighbors located 63° above the frontal plane of the body and 12° left and right of the sagittal plane of each eye; whereas for golden shiners, they may prefer positions where conspecifics are aligned 32° above the frontal plane of the body and 67° left and right of the sagittal plane for the left and right eye (Fig. 5A and 5B). Actually, there is evidence that information flows unidirectionally through fish groups such that individuals receive information and base their movement decisions from neighbors that are directly in front of them (Herbert-Read et al., 2011). Comparing these estimates with the literature, golden shiners have been shown to respond to changes in the directional movement of neighbors that are ahead of them rather than behind (Katz et al., 2011). Assuming that individuals use the center of acute vision to obtain social cues, and considering the between-species differences in the positioning of the areae, we can make predictions about the optimal position of individuals in a group from a visual sensory perspective (Figs. 5C and 5D). Taking into consideration the roughness of the school (surface area/volume ratio), we can predict the zebrafish will have a higher roughness value (i.e., more laterally compressed school; Fig. 5C) than the golden shiner (i.e., more vertically compressed school; Fig. 5D). Variations in roughness have been associated with access to oxygen and predator avoidance (Brierley & Cox, 2010). Unfortunately, there are no estimates of roughness for zebrafish and golden shiners in the literature.

Individuals often maintain a particular distance from group mates (i.e., nearest neighbor distance) (Krause & Ruxton, 2002) by using reference markings on the body (i.e., schooling marks, Bone & Moore, 2008). Based on our estimates of spatial resolving power, we calculated the maximum distance range that both cyprinid species would need to maintain in order to resolve visually these species-specific conspicuous markings, and thus remain attached to the group. In zebrafish, the centers of acute vision are capable of resolving the conspecific striping pattern up to 19.49 cm away using the fronto-dorsal region of their visual field, while the retinal periphery up to 8.25 cm away. Consequently, zebrafish would need to maintain a neighbor distance range of approximately 8–20 cm (∼2–5 body lengths) (Fig. 5A). Golden shiners could resolve the cycloid scales up to 19.66 cm away with their centers of acute vision that project to the fronto-dorsal region of the visual field and up to 8.11 cm with the retinal periphery. Therefore, golden shiners would need to maintain a neighbor separation range of also 8–20 cm (∼1–3 body lengths) (Fig. 5B). However, this distance range assumes ideal, full spectrum light conditions, maximum contrast between the conspecific and the background, and does not account for variations in the optical clarity of the water environment; all these factors could potentially shorten this distance. However, considering that both species tend to inhabit shallow, relatively clear waters (i.e., depths up to 30 cm for the zebrafish and <55 cm for the golden shiner), there is likely to be a low degree of attenuation and consequently, a wide range of available wavelengths to perceive conspecifics. Therefore, the interaction ranges provided by our calculations should not be constrained by the depth of the water environment (Sigler & Sigler, 1987; Krause, Godin & Brown, 1996; Spence et al., 2008). Comparing these estimates with the literature, lab experiments conducted on both species suggest that zebrafish maintain an average nearest neighbor separation of approximately 0.5–5.5 cm (∼1 body length), and golden shiners of 9–20 cm (∼1–3 body lengths) (Burgess & Shaw, 1981; Miller & Gerlai, 2012). Although our predicted distance ranges for the resolution of an entire conspecific are well above the behaviorally determined nearest neighbor distances of both species, our ranges based on the resolution of the conspicuous markings are more consistent with their natural behavior, supporting their role as a potential cue used during collective interactions. However, there are alternative explanations to the observed neighbor distances related to hydrodynamic efficiency (Hemelrijk & Hildenbrandt, 2012) and the use of vibrational cues to sense conspecifics via the lateral line (Partridge & Pitcher, 1980).

Concluding remarks

In the present study, we developed novel quantitative predictions for key parameters (position, spacing, orientation) underlying social interactions of zebrafish and golden shiners based on properties of their visual system. Therefore, an understanding of the species-specific features of the visual system allowed us to predict how individuals would position themselves in groups to optimize the transfer of visual social information. These predictions could be tested empirically in the future.

Most importantly, our findings contradict previous modeling assumptions, which assume that fish have a 60° blind zone and weigh information equally from multiple surrounding neighbors of a fixed distance. We found that the blind areas of both species were less than half the assumed width used in many models simulating fish. We also suggest that the maximum distance at which individuals are able to perceive conspecific cues varies according to different regions of the visual field: 2–5 body lengths in the zebrafish and 1–3 body lengths in the golden shiner, with frontal areas providing the longest distances. Additionally, considering our estimate is in accordance with both species’ behaviorally measured nearest neighbor distances (i.e., ∼1 body length of the zebrafish and 1–3 body lengths for the golden shiner) (Burgess & Shaw, 1981; Miller & Gerlai, 2012), it is possible that these species are using cues about changes in conspecific movement through the resolution of the conspicuous markings. It should also be noted that our distance range estimates for the resolution of another conspecific (i.e., 8–18 body lengths for the zebrafish and 16–39 body lengths for the golden shiner) were much higher than the behaviorally determined interaction range (0.5–2.0 body lengths) typically used in individual-based models to specify when an individual becomes aware of another fish. Thus, making strict sensory assumptions in collective behavior models may misrepresent the basis of local interactions and emerging group structure. The overall implication is that the sensory dimensions of a given species are likely to influence the local interactions driving its collective behavior. We suggest that to better understand the mechanisms driving local interactions in groups, more effort should be invested in assessing how the sensory system may shape the spatial ability to gather information from group mates.