Neuroscience 296 (2015) 75–79
REVIEW USING RATS FOR VISION RESEARCH P. REINAGEL *
powerful methods are difficult or costly to perform in animals with such large brains, however, including lesions, slice physiology, recording or imaging simultaneously from multiple visual areas, filling and reconstructing individual cells, anatomic tracing of longrange projections, genetic labeling of specific cell types, isolating mutants, transgenic animals, or comparisons across large populations of individuals. Accordingly, these methods are rarely employed in macaque vision research, leaving gaps in our knowledge. Therefore a small mammal with robust visual behavior would be valuable for our research and for the field in general.
Section of Neurobiology, Division of Biological Sciences, University of California, San Diego, La Jolla, CA, United States
Abstract—A wide variety of species are used for the study of visual neuroscience. This is beneficial because fundamental mechanisms and theoretical principles of vision are likely to be highly conserved, while different species exhibit different visual capacities and present different technical advantages for experiments. Eight years ago my laboratory adopted the hooded rat as our primary preparation for vision research. To some this may be surprising, as nocturnal rodents have often been presumed to have poor vision and weak visual behavior. This commentary will provide my personal perspective on how I came to work with rats; discuss an example research project for which rats have been advantageous; and comment on the opportunities and challenges of the preparation. This article is part of a Special Issue entitled: Contributions From Different Model Organisms to Brain Research. Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved.
HOW WE CHOSE RATS We considered a wide range of small mammal species, and commenced pilot studies of visual behavior of candidate species until a suitable preparation was identified. In choosing candidates we considered visual characteristics (prominence of visual system; percent of photoreceptors that are cones), experimental accessibility (suitability for optical imaging, slice preparation, behavioral tractability), pragmatic factors (size, availability, cost, maintenance burden, epizootic problems, ease of handling), and research infrastructure (existence of breeding colony, stereotaxic atlas, genome sequence, genetic libraries), to the extent that these facts were known. No species was ideal in all respects. Balancing the advantages and disadvantages in different ways, our top candidates were California ground squirrel, degu, gerbil, guinea pig, rat, and mouse. Our short list also included thirteen-lined ground squirrel, Nile rat, hamster, tree shrew, ferret, and bush baby. We began by testing four candidates. Two were chosen for highly developed, cone-dominated visual systems, in spite of limited research infrastructure: California ground squirrel (Otospermophilus beecheyi) and degu (Octodon degus). Two were chosen for highly developed infrastructure, in spite of limited visual systems: rat (Rattus norvegicus, specifically the pigmented Long-Evans strain), and mouse (Mus musculus, specifically the F1 hybrid of c57bl/6 dba/2). The goal was to find one small mammal that could learn a visual task by operant conditioning with twoalternative forced choice (2AFC) trial design (rather than go–nogo), appetitive reward (as against negative reinforcement, e.g., escaping water), liquid reward (as against food pellets), and computer-displayed visual
Key words: rodent vision, visual behavior, pigmented rat, visual psychophysics. Contents Introduction How we chose rats An example study using rats Strengths and limitations of the rat preparation Acknowledgements References Appendix A. Supplementary data
75 75 76 77 78 78 79
INTRODUCTION The development of the visually behaving macaque model revolutionized visual neuroscience, allowing neurophysiology to be directly linked with behavior on a trial-by-trial basis. Immense progress has been made in this preparation, and we expect it will remain the best or only preparation for many visual studies. Several *Address: Section of Neurobiology, Division of Biology, University of California, San Diego, 9500 Gilman Drive #0357, La Jolla, CA 92093, United States. E-mail address: [email protected] Abbreviation: 2AFC, two-alternative forced choice. http://dx.doi.org/10.1016/j.neuroscience.2014.12.025 0306-4522/Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved. 75
76
P. Reinagel / Neuroscience 296 (2015) 75–79
stimuli (as against physical objects or cue cards). In these respects, we mimicked conditions typical of human and monkey visual behavior experiments, which are well suited to providing a large number of trials in rapid succession with temporal precision and minimal human intervention. Our pilot task was discrimination of high-contrast lowspatial frequency sinusoidal gratings at mesopic mean luminance. We used an operant chamber that had been previously developed for 2AFC olfactory and auditory tasks in freely behaving rats (Uchida and Mainen, 2003; Hroma´dka and Zador, 2007; Otazu et al., 2009). For control of visual stimuli, trials, and training we used an early prototype of the training protocols and custom software that were further developed and described in detail elsewhere (Meier et al., 2011, and supplementary materials thereof). In the pilot study trials were initiated by insertion of nose in the center nose poke, and terminated by response at either the left or right nose poke. Correct responses earned liquid reward; incorrect responses earned a brief penalty time-out during which a flickering checkerboard was displayed (see Supplemental Videos 1 and 2). The first species we succeeded in training were LongEvans hooded rats (Fig. 1A, B; Supplementary Video 1) and California Ground Squirrels (Fig. 1C, D; Supplementary Video 2). The rats we tested were young adults (P30–P90). The squirrels we tested were captive-raised 1-year-old adults, born of wild-caught pregnant mothers. The epizootic risks necessitated BL2 handling, off-site housing, and daily transportation by car to the testing laboratory. Despite these sub-optimal conditions the squirrels learned rapidly and performed well in the task. In our initial pilot, neither mice nor degus learned the visual task. We chose to proceed
with rats rather than squirrels because rats were able to learn and perform all the visual tasks we needed, and the existing research infrastructure for rats is far better than for squirrels. With further effort, protocols have since been optimized for freely behaving mice as well (for preliminary report, see Sriram et al., 2013). We abandoned efforts to optimize protocols for degus, and did not test the other candidate species, so their visual and behavioral capacities in our task remain unknown. Using rats as our primary preparation, we developed software, hardware, and operant chambers for automated training and testing of rodents in visual tasks (Meier et al., 2011). The ability to train a large number of subjects in a compact space is one significant advantage of using a small mammal model. We note that other groups had previously (Birch and Jacobs, 1979; Cowey and Franzini, 1979; Keller et al., 2000; Prusky et al., 2000) or contemporaneously (Douglas et al., 2006; Minini and Jeffery, 2006; Bussey et al., 2008; Zoccolan et al., 2009) developed related methods for training and testing vision in rodents, and new methods are emerging daily, especially for mice (e.g., Chen et al., 2008; Harvey et al., 2009; Andermann et al., 2010; Dombeck et al., 2010; Niell and Stryker, 2010; Busse et al., 2011; Harvey et al., 2012; Histed et al., 2012).
AN EXAMPLE STUDY USING RATS We find rats to be excellent subjects for studies that require complex visual behaviors and large numbers of trials from multiple subjects. To illustrate this, consider our use of rats to study spatial context effects (Meier et al., 2011; Meier and Reinagel 2011; Meier and Reinagel 2013). A question of broad interest in vision research is to understand how spatially proximal visual
Fig. 1. Small animals tested for visual behavior. (A) Long-Evans hooded rat in pilot test, performing 2AFC orientation discrimination. (B) Performance of one of the rats as a function of spatial frequency (8437 trials over 18 days). (C) California Ground Squirrel in pilot test, performing 2AFC orientation discrimination. (D) Performance of one of the squirrels as a function of spatial frequency (797 trials over 2 days). Mice and Octodon degus were also tested, but performed poorly under the pilot test conditions.
P. Reinagel / Neuroscience 296 (2015) 75–79
information alters the primary perception of local visual cues. At a basic level, the effective contrast of local visual stimuli is normalized by surrounding contrast, a computation that arises as early as the retina. We were specifically interested in contextual interactions that are sensitive to higher organizing principles, such as the collinearity of an oriented stimulus with its surrounding context. It is plausible that specific lateral interactions in primary visual cortex might underlie such configuration-dependent perceptual effects. This hypothesis has been difficult to test, however, as the behavioral evidence has been mainly from humans. Therefore our goal was to establish a small animal model for configuration-sensitive contextual effects on visual perception. The task we adopted is a flanker detection task. The subject must report the presence or absence of a small oriented grating patch (target) in the center of the display, while ignoring oriented grating patches at other nearby locations flanking the target (Fig. 2A). In humans, detection is affected by flanking stimuli in general, and by collinear flankers in particular (Ejima and Takahashi, 1985; Polat and Sagi, 1993, 2007; Xing and Heeger, 2001; Chen and Tyler, 2008; Chubb et al., 1989; Zenger and Sagi, 1996; Cannon and Fullenkamp, 1996; Williams and Hess, 1998; Solomon and Morgan, 2000). Flanker tasks have also been explored in nonhuman primates (Li et al., 2006; Pooresmaeili et al., 2010). There were many challenges and unknowns in adapting this task to rats. It was unclear whether rats would be able to learn such a task, which demands a kind of spatially selective attention. Moreover, the effects of context can be small, sometimes just a few percent change in performance. To resolve such subtle effects, rats would need to perform a large number of trials with stable performance. Even if effects such as contrast normalization could be demonstrated in rats, rodents lack orientation-column architecture in V1, which might have been essential for configurationspecific lateral interactions. We found, however, that rats could perform the task, contextual effects exist and could be resolved, and the effect of flankers was sensitive to collinear configuration (Fig. 2B, C; Meier et al., 2011). The advantage of using rats for this study was the ability to train several (N = 7) subjects in parallel on a difficult visual task and collect very large numbers of trials in a modest time frame. On average, naı¨ ve rats reached stable performance on the flanker detection task in a total of 26,000 trials over 103 days. The rats then had stable performance for months, allowing us to collect on average 55,000 trials of behavioral data per rat over a period of 2–5 months (98 ± 63 days). This large trial number provided the statistical power to fully explore relevant stimulus parameters and control conditions, and to resolve even small differences in performance across conditions. Because all seven subjects were trained in parallel, the entire training and data collection period was about six months. Rats have a natural lifespan of at least two years in the laboratory, allowing highly trained subjects to be used for multiple
77
Fig. 2. Rats tested in a flanker detection task. (A) Rat performing the task, which was to report the presence or absence of a central grating patch while ignoring the flanking patches. (B) Average performance (% correct) of rats as a function of flanker configuration. When flankers were collinear (red), detection was selectively impaired (N = 7 rats). (C) Difference in performance between the collinear condition and a configuration with non-collinear flankers at the same locations. For six individual rats (blue), performance was significantly lower when flankers were collinear. For one rat the difference was not significant (gray). (Image and figures reproduced from Meier et al., 2011).
experiments. In this study, for example, two rats were tested for another three months while we varied a different set of stimulus parameters, allowing us to build a detailed quantitative model that was highly constrained by empirical measurements (Meier and Reinagel, 2011). We are not aware of any other animal model in which one could produce such a large behavioral dataset in an equivalent time frame and budget. There are two significant caveats to this story. First, we have only behavioral data so far, which could also have been collected from human subjects. The power of having an animal model is yet to be exploited. But in principle, manipulation of the visual circuits during behavior is now feasible. Moreover, our quantitative models can be usefully combined with independently collected data on the biophysics, cellular physiology, micro circuitry, and visual electrophysiology of rat primary visual cortex. The second caveat is that the psychophysical effects we find in rats are not the same as found in humans (Meier and Reinagel, 2013). Therefore the rat is not an exact model for the visual processing and sensory decision-making of humans. What the rat does provide is a viable model of how configuration-sensitive spatial context effects, including collinearity computation, can arise in a mammalian visual cortex.
STRENGTHS AND LIMITATIONS OF THE RAT PREPARATION Based on our work and work in other labs, we now know that rats can rapidly learn and reliably perform many complex visual tasks, such as visual object discrimination (Zoccolan et al., 2009; Clark et al., 2011; Tafazoli et al., 2012; Vermaercke and Op de Beeck,
78
P. Reinagel / Neuroscience 296 (2015) 75–79
2012; Alemi-Neissi et al., 2013; Petruno et al., 2013; Brooks et al., 2013), random dot motion discrimination (Douglas et al., 2006; Petruno et al. 2013; Reinagel, 2013), and detection with flanking distractors (Meier et al., 2011; Meier and Reinagel, 2011). The flanker detection is the most difficult task we have trained rats to perform, as judged by the unusually long training time (100 days). Other visual tasks can be acquired much more quickly. For example, if only detection is required, rats can learn to report the side on which a stimulus appears in just a few days, and to perform 2-AFC visual discrimination in about a week (Clark et al., 2011; Petruno et al., 2013; Reinagel 2013). In addition to visual psychophysics, we also find the rat preparation valuable for studies requiring complex visual behavior in combination with lesions. For example, rats can perform subtle visual object discriminations, which enabled a test of the role of the perirhinal cortex in this visual computation (Clark et al., 2011). Individual rats can be trained to perform a battery of disparate tasks (stimulus localization, orientation discrimination, random-dot motion discrimination, and photographic image discrimination); this has enabled a test of the perceptual function of V1 and medial extra-striate cortex (Petruno et al., 2013). Primates could easily perform the same behaviors, but lesion studies in primates are costly and necessarily limited in number. Although mice are amenable to lesions, we have so far found them more difficult to train. An important motivation for using rats is that we are developing behavioral paradigms and computational theories in a preparation that will also be accessible to tools of circuit analysis and manipulation going forward. Compared with the behaving macaque monkey, we expect that visually behaving rats will be a preferred preparation for methods such as slice physiology, viral vector gene expression, cell filling, and anatomic projection tracing. It remains to be determined whether rats can perform all the complex visual tasks now established in the macaque preparation, however. Rats have very few cones, lack color vision, and have a very limited binocular field (although California Ground Squirrels are comparable to macaques in these respects). For studies requiring sophisticated genetics, the mouse will be the preparation of choice, although it remains to be determined whether mice can learn and perform all the tasks that are now established in rats. In summary, we find the rat to be a valuable preparation for vision research that is complementary to other established and emerging models. Acknowledgements—The opinions expressed here are my own. Without implication of endorsement, I wish to acknowledge my debt to Harvey Karten for valuable discussion in the identification of candidate species, and to Tony Zador and Zach Mainen for helping us replicate and adapt their rat behavioral apparatus. The behavioral pilot experiments were carried out in my laboratory by Erik Flister, Philip Meier, Bevil Conway, and Elizabeth Murphy, with expert technical support from Pamela SchupMagoffin. The multi-species pilot study was supported by the Hellman Fellows Fund.
REFERENCES Alemi-Neissi A, Rosselli FB, Zoccolan D (2013) Multifeatural shape processing in rats engaged in invariant visual object recognition. J Neurosci 33:5939–5956. Andermann ML, Kerlin AM, Reid RC (2010) Chronic cellular imaging of mouse visual cortex during operant behavior and passive viewing Front. Cell Neurosci 4:3. Birch D, Jacobs GH (1979) Spatial contrast sensitivity in albino and pigmented rats. Vis Res 19:933–937. Brooks DI, Ng KH, Buss EW, Marshall AT, Freeman JH, Wasserman EA (2013) Categorization of photographic images by rats using shape-based image dimensions. J Exp Psychol Anim Behav Process 39(1):85–92. Busse L, Ayaz A, Dhruv NT, Katzner S, Saleem AB, Scholvinck ML, Zaharia AD, Carandini M (2011) The detection of visual contrast in the behaving mouse. J Neurosci 31:11351–11361. Bussey TJ, Padain TL, Skillings EA, Winters BD, Morton AJ, Saksida LM (2008) The touch screen cognitive testing method for rodents: how to get the best out of your rat. Learn Membr 15:516–523. Cannon MW, Fullenkamp SC (1996) A model for inhibitory lateral interaction effects in perceived contrast. Vis Res 36:1115–1125. Chen CC, Tyler CW (2008) Excitatory and inhibitory interaction fields of flankers revealed by contrast-masking functions. J Vis 8(4):10.1–10.14. Chen Y, Martinez-Conde S, Macknik SL, Bereshpolova Y, Swadlow HA, Alonso JM (2008) Task difficulty modulates the activity of specific neuronal populations in primary visual cortex. Nat Neurosci 11:974–982. Chubb C, Sperling G, Solomon JA (1989) Texture interactions determine perceived contrast. Proc Natl Acad Sci U S A 86:9631–9635. Clark RE, Reinagel P, Broadbent NJ, Flister ED, Squire LR (2011) Intact performance on feature ambiguous discriminations in rats with lesions of the perirhinal cortex. Neuron 70(1):132–140. Cowey A, Franzini C (1979) The retinal origin of uncrossed optic nerve fibres in rats and their role in visual discrimination. Exp Brain Res 35:443–455. Dombeck DA, Harvey CD, Tian L, Looger LL, Tank DW (2010) Functional imaging of hippocampal place cells at cellular resolution during virtual navigation. Nat Neurosci 13:1433–1440. Douglas RM, Neve A, Quittenbaum JP, Alam NM, Prusky GT (2006) Perception of visual motion coherence by rats and mice. Vis Res 46:2842–2847. Ejima Y, Takahashi S (1985) Apparent contrast of a sinusoidal grating in the simultaneous presence of peripheral gratings. Vis Res 25:1223–1232. Harvey CD, Coen P, Tank DW (2012) Choice-specific sequences in parietal cortex during a virtual-navigation decision task. Nature 484:62–68. Harvey CD, Collman F, Dombeck DA, Tank DW (2009) Intracellular dynamics of hippocampal place cells during virtual navigation. Nature 461(7266):941–946. Histed MH, Carvalho LA, Maunsell JH (2012) Psychophysical measurement of contrast sensitivity in the behaving mouse. J Neurophysiol 107:758–765. Hroma´dka T, Zador AM (2007) Toward the mechanisms of auditory attention. Hear Res 229:180–185. Li W, Piech V, Gilbert CD (2006) Contour saliency in primary visual cortex. Neuron 50:951–962. Keller J, Strasburger H, Cerutti DT, Sabel BA (2000) Assessing spatial vision: automated measurement of the contrast-sensitivity function in the hooded rat. J Neurosci Methods 97:103–110. Meier PM, Reinagel P (2011) Rat performance on visual detection task modeled with divisive normalization and adaptive decision thresholds. J Vis 11(9). article 1. Meier P, Reinagel P (2013) Rats and humans differ in processing collinear visual features. Front Neural Circuits 7:197. Meier PM, Flister ED, Reinagel P (2011) Collinear features impair visual detection by rats. J Vis 11(3). article 22.
P. Reinagel / Neuroscience 296 (2015) 75–79 Minini L, Jeffery KJ (2006) Do rats use shape to solve ‘‘shape discriminations’’? Learn Membr 13:287–297. Niell CM, Stryker MP (2010) Modulation of visual responses by behavioral state in mouse visual cortex. Neuron 65:472–479. Otazu GH, Tai LH, Yang Y, Zador AM (2009) Engaging in an auditory task suppresses responses in auditory cortex. Nat Neurosci 12:646–654. Petruno S, Clark RE, Reinagel P (2013) Evidence that primary visual cortex is required for image orientation and motion discrimination by rats. PLoS One 8(2):e56543. Polat U, Sagi D (1993) Lateral interactions between spatial channels: suppression and facilitation revealed by lateral masking experiments. Vis Res 33:993–999. Polat U, Sagi D (2007) The relationship between the subjective and objective aspects of visual filling-in. Vis Res 47:2473–2481. Pooresmaeili A, Herrero JL, Self MW, Roelfsema PR, Thiele A (2010) Suppressive lateral interactions at parafoveal representations in primary visual cortex. J Neurosci 30(38):12745–12758. Prusky GT, West PWR, Douglas RM (2000) Behavioral assessment of visual acuity in mice and rats. Vis Res 40:2201–2209. Reinagel P (2013) Speed and accuracy of visual motion discrimination by rats. PLoS One 8(6):e68505. Solomon JA, Morgan MJ (2000) Facilitation from collinear flanks is cancelled by non-collinear flanks. Vis Res 40:279–286. Sriram B, Cruz-Martin A, Denardo L, Kim EJ, Patel M, Huynh P, Giering E, Reinagel P, Ghosh, A. Visually guided behavior in freely moving mice. Program No. 825.12; 2013 Neuroscience Meeting Planner. San Diego, CA: Society for Neuroscience, 2013. Online.
79
Tafazoli S, DiFilippo A, Zoccolan D (2012) Transformation-tolerant object recognition in rats revealed by visual priming. J Neurosci 32:21–34. Uchida N, Mainen ZF (2003) Speed and accuracy of olfactory discrimination in the rat. Nat Neurosci 6(11):1224–1229. Vermaercke B, Op de Beeck HP (2012) A multivariate approach reveals the behavioral templates underlying visual discrimination in rats. Curr Biol 22(1):50–55. Williams CB, Hess RF (1998) Relationship between facilitation at threshold and suprathreshold contour integration. J Opt Soc Am A Op Image Sci Vis 15:2046–2051. Xing J, Heeger DJ (2001) Measurement and modeling of center– surround suppression and enhancement. Vis Res 41:571–583. Zenger B, Sagi D (1996) Isolating excitatory and inhibitory nonlinear spatial interactions involved in contrast detection. Vis Res 36:2497–2513. Zoccolan D, Oertelt N, DiCarlo JJ, Cox DD (2009) A rodent model for the study of invariant visual object recognition. Proc Natl Acad Sci U S A 106:8748–8753.
APPENDIX A. SUPPLEMENTARY DATA Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.neuroscience.2014.12.025.
(Accepted 13 December 2014) (Available online 24 December 2014)
REVIEW USING RATS FOR VISION RESEARCH P. REINAGEL *
powerful methods are difficult or costly to perform in animals with such large brains, however, including lesions, slice physiology, recording or imaging simultaneously from multiple visual areas, filling and reconstructing individual cells, anatomic tracing of longrange projections, genetic labeling of specific cell types, isolating mutants, transgenic animals, or comparisons across large populations of individuals. Accordingly, these methods are rarely employed in macaque vision research, leaving gaps in our knowledge. Therefore a small mammal with robust visual behavior would be valuable for our research and for the field in general.
Section of Neurobiology, Division of Biological Sciences, University of California, San Diego, La Jolla, CA, United States
Abstract—A wide variety of species are used for the study of visual neuroscience. This is beneficial because fundamental mechanisms and theoretical principles of vision are likely to be highly conserved, while different species exhibit different visual capacities and present different technical advantages for experiments. Eight years ago my laboratory adopted the hooded rat as our primary preparation for vision research. To some this may be surprising, as nocturnal rodents have often been presumed to have poor vision and weak visual behavior. This commentary will provide my personal perspective on how I came to work with rats; discuss an example research project for which rats have been advantageous; and comment on the opportunities and challenges of the preparation. This article is part of a Special Issue entitled: Contributions From Different Model Organisms to Brain Research. Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved.
HOW WE CHOSE RATS We considered a wide range of small mammal species, and commenced pilot studies of visual behavior of candidate species until a suitable preparation was identified. In choosing candidates we considered visual characteristics (prominence of visual system; percent of photoreceptors that are cones), experimental accessibility (suitability for optical imaging, slice preparation, behavioral tractability), pragmatic factors (size, availability, cost, maintenance burden, epizootic problems, ease of handling), and research infrastructure (existence of breeding colony, stereotaxic atlas, genome sequence, genetic libraries), to the extent that these facts were known. No species was ideal in all respects. Balancing the advantages and disadvantages in different ways, our top candidates were California ground squirrel, degu, gerbil, guinea pig, rat, and mouse. Our short list also included thirteen-lined ground squirrel, Nile rat, hamster, tree shrew, ferret, and bush baby. We began by testing four candidates. Two were chosen for highly developed, cone-dominated visual systems, in spite of limited research infrastructure: California ground squirrel (Otospermophilus beecheyi) and degu (Octodon degus). Two were chosen for highly developed infrastructure, in spite of limited visual systems: rat (Rattus norvegicus, specifically the pigmented Long-Evans strain), and mouse (Mus musculus, specifically the F1 hybrid of c57bl/6 dba/2). The goal was to find one small mammal that could learn a visual task by operant conditioning with twoalternative forced choice (2AFC) trial design (rather than go–nogo), appetitive reward (as against negative reinforcement, e.g., escaping water), liquid reward (as against food pellets), and computer-displayed visual
Key words: rodent vision, visual behavior, pigmented rat, visual psychophysics. Contents Introduction How we chose rats An example study using rats Strengths and limitations of the rat preparation Acknowledgements References Appendix A. Supplementary data
75 75 76 77 78 78 79
INTRODUCTION The development of the visually behaving macaque model revolutionized visual neuroscience, allowing neurophysiology to be directly linked with behavior on a trial-by-trial basis. Immense progress has been made in this preparation, and we expect it will remain the best or only preparation for many visual studies. Several *Address: Section of Neurobiology, Division of Biology, University of California, San Diego, 9500 Gilman Drive #0357, La Jolla, CA 92093, United States. E-mail address: [email protected] Abbreviation: 2AFC, two-alternative forced choice. http://dx.doi.org/10.1016/j.neuroscience.2014.12.025 0306-4522/Ó 2014 IBRO. Published by Elsevier Ltd. All rights reserved. 75
76
P. Reinagel / Neuroscience 296 (2015) 75–79
stimuli (as against physical objects or cue cards). In these respects, we mimicked conditions typical of human and monkey visual behavior experiments, which are well suited to providing a large number of trials in rapid succession with temporal precision and minimal human intervention. Our pilot task was discrimination of high-contrast lowspatial frequency sinusoidal gratings at mesopic mean luminance. We used an operant chamber that had been previously developed for 2AFC olfactory and auditory tasks in freely behaving rats (Uchida and Mainen, 2003; Hroma´dka and Zador, 2007; Otazu et al., 2009). For control of visual stimuli, trials, and training we used an early prototype of the training protocols and custom software that were further developed and described in detail elsewhere (Meier et al., 2011, and supplementary materials thereof). In the pilot study trials were initiated by insertion of nose in the center nose poke, and terminated by response at either the left or right nose poke. Correct responses earned liquid reward; incorrect responses earned a brief penalty time-out during which a flickering checkerboard was displayed (see Supplemental Videos 1 and 2). The first species we succeeded in training were LongEvans hooded rats (Fig. 1A, B; Supplementary Video 1) and California Ground Squirrels (Fig. 1C, D; Supplementary Video 2). The rats we tested were young adults (P30–P90). The squirrels we tested were captive-raised 1-year-old adults, born of wild-caught pregnant mothers. The epizootic risks necessitated BL2 handling, off-site housing, and daily transportation by car to the testing laboratory. Despite these sub-optimal conditions the squirrels learned rapidly and performed well in the task. In our initial pilot, neither mice nor degus learned the visual task. We chose to proceed
with rats rather than squirrels because rats were able to learn and perform all the visual tasks we needed, and the existing research infrastructure for rats is far better than for squirrels. With further effort, protocols have since been optimized for freely behaving mice as well (for preliminary report, see Sriram et al., 2013). We abandoned efforts to optimize protocols for degus, and did not test the other candidate species, so their visual and behavioral capacities in our task remain unknown. Using rats as our primary preparation, we developed software, hardware, and operant chambers for automated training and testing of rodents in visual tasks (Meier et al., 2011). The ability to train a large number of subjects in a compact space is one significant advantage of using a small mammal model. We note that other groups had previously (Birch and Jacobs, 1979; Cowey and Franzini, 1979; Keller et al., 2000; Prusky et al., 2000) or contemporaneously (Douglas et al., 2006; Minini and Jeffery, 2006; Bussey et al., 2008; Zoccolan et al., 2009) developed related methods for training and testing vision in rodents, and new methods are emerging daily, especially for mice (e.g., Chen et al., 2008; Harvey et al., 2009; Andermann et al., 2010; Dombeck et al., 2010; Niell and Stryker, 2010; Busse et al., 2011; Harvey et al., 2012; Histed et al., 2012).
AN EXAMPLE STUDY USING RATS We find rats to be excellent subjects for studies that require complex visual behaviors and large numbers of trials from multiple subjects. To illustrate this, consider our use of rats to study spatial context effects (Meier et al., 2011; Meier and Reinagel 2011; Meier and Reinagel 2013). A question of broad interest in vision research is to understand how spatially proximal visual
Fig. 1. Small animals tested for visual behavior. (A) Long-Evans hooded rat in pilot test, performing 2AFC orientation discrimination. (B) Performance of one of the rats as a function of spatial frequency (8437 trials over 18 days). (C) California Ground Squirrel in pilot test, performing 2AFC orientation discrimination. (D) Performance of one of the squirrels as a function of spatial frequency (797 trials over 2 days). Mice and Octodon degus were also tested, but performed poorly under the pilot test conditions.
P. Reinagel / Neuroscience 296 (2015) 75–79
information alters the primary perception of local visual cues. At a basic level, the effective contrast of local visual stimuli is normalized by surrounding contrast, a computation that arises as early as the retina. We were specifically interested in contextual interactions that are sensitive to higher organizing principles, such as the collinearity of an oriented stimulus with its surrounding context. It is plausible that specific lateral interactions in primary visual cortex might underlie such configuration-dependent perceptual effects. This hypothesis has been difficult to test, however, as the behavioral evidence has been mainly from humans. Therefore our goal was to establish a small animal model for configuration-sensitive contextual effects on visual perception. The task we adopted is a flanker detection task. The subject must report the presence or absence of a small oriented grating patch (target) in the center of the display, while ignoring oriented grating patches at other nearby locations flanking the target (Fig. 2A). In humans, detection is affected by flanking stimuli in general, and by collinear flankers in particular (Ejima and Takahashi, 1985; Polat and Sagi, 1993, 2007; Xing and Heeger, 2001; Chen and Tyler, 2008; Chubb et al., 1989; Zenger and Sagi, 1996; Cannon and Fullenkamp, 1996; Williams and Hess, 1998; Solomon and Morgan, 2000). Flanker tasks have also been explored in nonhuman primates (Li et al., 2006; Pooresmaeili et al., 2010). There were many challenges and unknowns in adapting this task to rats. It was unclear whether rats would be able to learn such a task, which demands a kind of spatially selective attention. Moreover, the effects of context can be small, sometimes just a few percent change in performance. To resolve such subtle effects, rats would need to perform a large number of trials with stable performance. Even if effects such as contrast normalization could be demonstrated in rats, rodents lack orientation-column architecture in V1, which might have been essential for configurationspecific lateral interactions. We found, however, that rats could perform the task, contextual effects exist and could be resolved, and the effect of flankers was sensitive to collinear configuration (Fig. 2B, C; Meier et al., 2011). The advantage of using rats for this study was the ability to train several (N = 7) subjects in parallel on a difficult visual task and collect very large numbers of trials in a modest time frame. On average, naı¨ ve rats reached stable performance on the flanker detection task in a total of 26,000 trials over 103 days. The rats then had stable performance for months, allowing us to collect on average 55,000 trials of behavioral data per rat over a period of 2–5 months (98 ± 63 days). This large trial number provided the statistical power to fully explore relevant stimulus parameters and control conditions, and to resolve even small differences in performance across conditions. Because all seven subjects were trained in parallel, the entire training and data collection period was about six months. Rats have a natural lifespan of at least two years in the laboratory, allowing highly trained subjects to be used for multiple
77
Fig. 2. Rats tested in a flanker detection task. (A) Rat performing the task, which was to report the presence or absence of a central grating patch while ignoring the flanking patches. (B) Average performance (% correct) of rats as a function of flanker configuration. When flankers were collinear (red), detection was selectively impaired (N = 7 rats). (C) Difference in performance between the collinear condition and a configuration with non-collinear flankers at the same locations. For six individual rats (blue), performance was significantly lower when flankers were collinear. For one rat the difference was not significant (gray). (Image and figures reproduced from Meier et al., 2011).
experiments. In this study, for example, two rats were tested for another three months while we varied a different set of stimulus parameters, allowing us to build a detailed quantitative model that was highly constrained by empirical measurements (Meier and Reinagel, 2011). We are not aware of any other animal model in which one could produce such a large behavioral dataset in an equivalent time frame and budget. There are two significant caveats to this story. First, we have only behavioral data so far, which could also have been collected from human subjects. The power of having an animal model is yet to be exploited. But in principle, manipulation of the visual circuits during behavior is now feasible. Moreover, our quantitative models can be usefully combined with independently collected data on the biophysics, cellular physiology, micro circuitry, and visual electrophysiology of rat primary visual cortex. The second caveat is that the psychophysical effects we find in rats are not the same as found in humans (Meier and Reinagel, 2013). Therefore the rat is not an exact model for the visual processing and sensory decision-making of humans. What the rat does provide is a viable model of how configuration-sensitive spatial context effects, including collinearity computation, can arise in a mammalian visual cortex.
STRENGTHS AND LIMITATIONS OF THE RAT PREPARATION Based on our work and work in other labs, we now know that rats can rapidly learn and reliably perform many complex visual tasks, such as visual object discrimination (Zoccolan et al., 2009; Clark et al., 2011; Tafazoli et al., 2012; Vermaercke and Op de Beeck,
78
P. Reinagel / Neuroscience 296 (2015) 75–79
2012; Alemi-Neissi et al., 2013; Petruno et al., 2013; Brooks et al., 2013), random dot motion discrimination (Douglas et al., 2006; Petruno et al. 2013; Reinagel, 2013), and detection with flanking distractors (Meier et al., 2011; Meier and Reinagel, 2011). The flanker detection is the most difficult task we have trained rats to perform, as judged by the unusually long training time (100 days). Other visual tasks can be acquired much more quickly. For example, if only detection is required, rats can learn to report the side on which a stimulus appears in just a few days, and to perform 2-AFC visual discrimination in about a week (Clark et al., 2011; Petruno et al., 2013; Reinagel 2013). In addition to visual psychophysics, we also find the rat preparation valuable for studies requiring complex visual behavior in combination with lesions. For example, rats can perform subtle visual object discriminations, which enabled a test of the role of the perirhinal cortex in this visual computation (Clark et al., 2011). Individual rats can be trained to perform a battery of disparate tasks (stimulus localization, orientation discrimination, random-dot motion discrimination, and photographic image discrimination); this has enabled a test of the perceptual function of V1 and medial extra-striate cortex (Petruno et al., 2013). Primates could easily perform the same behaviors, but lesion studies in primates are costly and necessarily limited in number. Although mice are amenable to lesions, we have so far found them more difficult to train. An important motivation for using rats is that we are developing behavioral paradigms and computational theories in a preparation that will also be accessible to tools of circuit analysis and manipulation going forward. Compared with the behaving macaque monkey, we expect that visually behaving rats will be a preferred preparation for methods such as slice physiology, viral vector gene expression, cell filling, and anatomic projection tracing. It remains to be determined whether rats can perform all the complex visual tasks now established in the macaque preparation, however. Rats have very few cones, lack color vision, and have a very limited binocular field (although California Ground Squirrels are comparable to macaques in these respects). For studies requiring sophisticated genetics, the mouse will be the preparation of choice, although it remains to be determined whether mice can learn and perform all the tasks that are now established in rats. In summary, we find the rat to be a valuable preparation for vision research that is complementary to other established and emerging models. Acknowledgements—The opinions expressed here are my own. Without implication of endorsement, I wish to acknowledge my debt to Harvey Karten for valuable discussion in the identification of candidate species, and to Tony Zador and Zach Mainen for helping us replicate and adapt their rat behavioral apparatus. The behavioral pilot experiments were carried out in my laboratory by Erik Flister, Philip Meier, Bevil Conway, and Elizabeth Murphy, with expert technical support from Pamela SchupMagoffin. The multi-species pilot study was supported by the Hellman Fellows Fund.
REFERENCES Alemi-Neissi A, Rosselli FB, Zoccolan D (2013) Multifeatural shape processing in rats engaged in invariant visual object recognition. J Neurosci 33:5939–5956. Andermann ML, Kerlin AM, Reid RC (2010) Chronic cellular imaging of mouse visual cortex during operant behavior and passive viewing Front. Cell Neurosci 4:3. Birch D, Jacobs GH (1979) Spatial contrast sensitivity in albino and pigmented rats. Vis Res 19:933–937. Brooks DI, Ng KH, Buss EW, Marshall AT, Freeman JH, Wasserman EA (2013) Categorization of photographic images by rats using shape-based image dimensions. J Exp Psychol Anim Behav Process 39(1):85–92. Busse L, Ayaz A, Dhruv NT, Katzner S, Saleem AB, Scholvinck ML, Zaharia AD, Carandini M (2011) The detection of visual contrast in the behaving mouse. J Neurosci 31:11351–11361. Bussey TJ, Padain TL, Skillings EA, Winters BD, Morton AJ, Saksida LM (2008) The touch screen cognitive testing method for rodents: how to get the best out of your rat. Learn Membr 15:516–523. Cannon MW, Fullenkamp SC (1996) A model for inhibitory lateral interaction effects in perceived contrast. Vis Res 36:1115–1125. Chen CC, Tyler CW (2008) Excitatory and inhibitory interaction fields of flankers revealed by contrast-masking functions. J Vis 8(4):10.1–10.14. Chen Y, Martinez-Conde S, Macknik SL, Bereshpolova Y, Swadlow HA, Alonso JM (2008) Task difficulty modulates the activity of specific neuronal populations in primary visual cortex. Nat Neurosci 11:974–982. Chubb C, Sperling G, Solomon JA (1989) Texture interactions determine perceived contrast. Proc Natl Acad Sci U S A 86:9631–9635. Clark RE, Reinagel P, Broadbent NJ, Flister ED, Squire LR (2011) Intact performance on feature ambiguous discriminations in rats with lesions of the perirhinal cortex. Neuron 70(1):132–140. Cowey A, Franzini C (1979) The retinal origin of uncrossed optic nerve fibres in rats and their role in visual discrimination. Exp Brain Res 35:443–455. Dombeck DA, Harvey CD, Tian L, Looger LL, Tank DW (2010) Functional imaging of hippocampal place cells at cellular resolution during virtual navigation. Nat Neurosci 13:1433–1440. Douglas RM, Neve A, Quittenbaum JP, Alam NM, Prusky GT (2006) Perception of visual motion coherence by rats and mice. Vis Res 46:2842–2847. Ejima Y, Takahashi S (1985) Apparent contrast of a sinusoidal grating in the simultaneous presence of peripheral gratings. Vis Res 25:1223–1232. Harvey CD, Coen P, Tank DW (2012) Choice-specific sequences in parietal cortex during a virtual-navigation decision task. Nature 484:62–68. Harvey CD, Collman F, Dombeck DA, Tank DW (2009) Intracellular dynamics of hippocampal place cells during virtual navigation. Nature 461(7266):941–946. Histed MH, Carvalho LA, Maunsell JH (2012) Psychophysical measurement of contrast sensitivity in the behaving mouse. J Neurophysiol 107:758–765. Hroma´dka T, Zador AM (2007) Toward the mechanisms of auditory attention. Hear Res 229:180–185. Li W, Piech V, Gilbert CD (2006) Contour saliency in primary visual cortex. Neuron 50:951–962. Keller J, Strasburger H, Cerutti DT, Sabel BA (2000) Assessing spatial vision: automated measurement of the contrast-sensitivity function in the hooded rat. J Neurosci Methods 97:103–110. Meier PM, Reinagel P (2011) Rat performance on visual detection task modeled with divisive normalization and adaptive decision thresholds. J Vis 11(9). article 1. Meier P, Reinagel P (2013) Rats and humans differ in processing collinear visual features. Front Neural Circuits 7:197. Meier PM, Flister ED, Reinagel P (2011) Collinear features impair visual detection by rats. J Vis 11(3). article 22.
P. Reinagel / Neuroscience 296 (2015) 75–79 Minini L, Jeffery KJ (2006) Do rats use shape to solve ‘‘shape discriminations’’? Learn Membr 13:287–297. Niell CM, Stryker MP (2010) Modulation of visual responses by behavioral state in mouse visual cortex. Neuron 65:472–479. Otazu GH, Tai LH, Yang Y, Zador AM (2009) Engaging in an auditory task suppresses responses in auditory cortex. Nat Neurosci 12:646–654. Petruno S, Clark RE, Reinagel P (2013) Evidence that primary visual cortex is required for image orientation and motion discrimination by rats. PLoS One 8(2):e56543. Polat U, Sagi D (1993) Lateral interactions between spatial channels: suppression and facilitation revealed by lateral masking experiments. Vis Res 33:993–999. Polat U, Sagi D (2007) The relationship between the subjective and objective aspects of visual filling-in. Vis Res 47:2473–2481. Pooresmaeili A, Herrero JL, Self MW, Roelfsema PR, Thiele A (2010) Suppressive lateral interactions at parafoveal representations in primary visual cortex. J Neurosci 30(38):12745–12758. Prusky GT, West PWR, Douglas RM (2000) Behavioral assessment of visual acuity in mice and rats. Vis Res 40:2201–2209. Reinagel P (2013) Speed and accuracy of visual motion discrimination by rats. PLoS One 8(6):e68505. Solomon JA, Morgan MJ (2000) Facilitation from collinear flanks is cancelled by non-collinear flanks. Vis Res 40:279–286. Sriram B, Cruz-Martin A, Denardo L, Kim EJ, Patel M, Huynh P, Giering E, Reinagel P, Ghosh, A. Visually guided behavior in freely moving mice. Program No. 825.12; 2013 Neuroscience Meeting Planner. San Diego, CA: Society for Neuroscience, 2013. Online.
79
Tafazoli S, DiFilippo A, Zoccolan D (2012) Transformation-tolerant object recognition in rats revealed by visual priming. J Neurosci 32:21–34. Uchida N, Mainen ZF (2003) Speed and accuracy of olfactory discrimination in the rat. Nat Neurosci 6(11):1224–1229. Vermaercke B, Op de Beeck HP (2012) A multivariate approach reveals the behavioral templates underlying visual discrimination in rats. Curr Biol 22(1):50–55. Williams CB, Hess RF (1998) Relationship between facilitation at threshold and suprathreshold contour integration. J Opt Soc Am A Op Image Sci Vis 15:2046–2051. Xing J, Heeger DJ (2001) Measurement and modeling of center– surround suppression and enhancement. Vis Res 41:571–583. Zenger B, Sagi D (1996) Isolating excitatory and inhibitory nonlinear spatial interactions involved in contrast detection. Vis Res 36:2497–2513. Zoccolan D, Oertelt N, DiCarlo JJ, Cox DD (2009) A rodent model for the study of invariant visual object recognition. Proc Natl Acad Sci U S A 106:8748–8753.
APPENDIX A. SUPPLEMENTARY DATA Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.neuroscience.2014.12.025.
(Accepted 13 December 2014) (Available online 24 December 2014)
