ESSAY EPPENDORF
Space Bats: Multidimensional Spatial Representation in the Bat
A novel animal model, the bat, is used to elucidate place-cell and grid-cell phenomena.
Michael M. Yartsev
Magnitude
(Right) Power-spectrum of a single grid-cell spike-train autocorrelogram showing no power in the theta frequency range (5 to 11 Hz, highlighted in red). (B) (Left) Illustration of a bat flying with the custom-made telemetry system, drawn to scale Princeton Neuroscience Institute, Princ- [Illustration: S. Kaufman]. (Middle) Color-coded three-dimensional firing-rate map, with peak rate indicated, of a single eton University, Princeton, NJ 08540, USA. 3D place cell recorded from the hippocampus of a freely flying bat. (Right) All place fields (individual colored convex hulls) E-mail: [email protected] recorded from the hippocampus of a single bat. Figures in (A) and (B) are adopted from (8) and (14), respectively.
www.sciencemag.org
SCIENCE
VOL 342
Published by AAAS
1 NOVEMBER 2013
573
Downloaded from www.sciencemag.org on March 17, 2015
I
t is estimated that more than 8 million tinuous theta oscillations (7, 8), perhaps different species reside on our planet, grid cells might also exist in their absence. many of which live a very different lifeThis would causally argue against the validstyle (1). But whether on the ground, in the ity of the oscillatory-interference class of ocean depths, or in the sky, all animals must models or, at least, against their generality have knowledge of their whereabouts to suracross mammals. We recorded the activvive. The possible mechanisms subserving ity of single MEC neurons in bats crawling this core function and how these are imple- Eppendorf and Science are pleased to present the inside a large arena (8) and found many of mented in the brain have been long-standing prize-winning essay by Michael M. Yartsev, the 2013 them to be grid cells [see the figure (A), left] questions in neuroscience. winner of the Eppendorf and Science Prize for Neu- with properties strikingly similar to those Scientists began studying the neural basis robiology. previously described in rats. We further of spatial representation with the report of found in the bat MEC all the other spatial neurons in the rat dorsal hippocampus that to the grid formation. Two major classes cell types previously described in the rat fired when the animal entered a specific loca- of computational models were proposed to MEC, such as neurons that encode the anition (2). These neurons were called “place account for this phenomena (6): Attractor- mal’s head orientation (9) and the borders cells.” More recently, scientists discovered based network models and single-neuron, of its current environment (10). We even “grid-cells” in the rat medial entorhinal “oscillatory-interference” models relying found many of the same neural oscillations cortex (MEC), which activate in multiple on theta-band (5 to 11 Hz) oscillations. The previously reported in the rat, such as highlocations, all arranged on the vertexes of a latter class of models received much exper- frequency “ripple” oscillations (11) and fast hexagonal grid (3, 4). Place cells and grid imental attention in rodent studies, but all and slow gamma oscillations (12). However, cells are widely considered key elements of evidences were of a correlative nature (6). as we had hypothesized, theta oscillations the mammalian spatial representation sys- We reasoned that, because place cells in bat in the bat were very different from those in tem (5), yet their detailed properties have hippocampus exist in the absence of con- rats. Theta oscillations in the bat were not been studied almost exclusively in rodents. This convergence in 1.6 Hz Theta choice of an animal model has A occluded the understanding of which neuronal mechanisms involved in spatial representation generalize across species and whether different solutions have been reached by different brains. To explore this ques0 10 20 30 tion and to study the functional Frequency (Hz) properties of these cell types, we B 15 Hz used a novel mammalian animal model, the bat. The findings in the bat helped revise our understanding of this important circuitry and show how divergence can powerfully complement convergence in the choice of animal models in neuroscience. We first asked whether the Properties of spatial cell types in crawling and flying bats. (A) (Left) Color-coded two-dimensional firing-rate map, with bat can provide insight into the peak rate indicated, of a single grid cell recorded from the MEC of a crawling bat. (Middle) Raw local field potential trace recorded from the MEC of a crawling bat showing a single intermittent theta-bout (top: 20-s trace; bottom: 7-s zoom-in). neural mechanisms giving rise
ESSAY
574
2013 Grand Prize Winner The author of the prize-winning essay, Michael Yartsev, received his undergraduate and master’s degrees in biomedical engineering from Ben-Gurion University in 2007. For his Ph.D., he joined the lab of Dr. Nachum Ulanovsky at the Weizmann Institute of Science. There, he recorded the activity of single neurons from the hippocampal formation of freely behaving and flying bats to study the underlying neural mechanisms of spatial memory and navigation in the mammalian brain. Since 2012, Dr. Yartsev is a C. V. Starr Fellow at the Princeton Neuroscience Institute at Princeton University where he is conducting postdoctoral work in the lab of Prof. Carlos Brody studying the neural basis of decision-making.
Finalists Daniel Bendor for his essay, “Play it again, brain.” Dr. Bendor is a lecturer in the Department of Cognitive, Perceptual, and Brain Sciences and the Institute of Behavioral Neuroscience at University College London. Dr. Bendor received his Ph.D. from Johns Hopkins University under the mentorship of Dr. Xiaoqin Wang, studying temporal processing in auditory cortex and the neural correlate of pitch and flutter perception. For his postdoctoral research, he investigated the role of the hippocampus in memory encoding and consolidation, while working with Dr. Matthew Wilson at the Massachusetts Institute of Technology. He has recently started his own lab at University College London, where his research focuses on how neural ensembles encode perceptual and memory-related information. http://scim.ag/_Bendor Sophie Caron for her essay, “Brains don’t play dice—or do they?” Dr. Caron is currently a postdoctoral fellow in the Department of Neuroscience at Columbia University. Sophie grew up in St-Blaise-sur-Richelieu in Canada and earned a B.Sc. in Biochemistry at the Université de Montréal. She moved to New York City to study the developmental mechanisms behind the diversification of sensory neurons in the laboratory of Dr. Alexander Schier at New York University and, later, Harvard. Having completed her Ph.D., Sophie joined the laboratory of Dr. Richard Axel at Columbia University, where she studies how the information gathered through the senses is represented in higher brain centers; in particular, those involved in memory. http://scim.ag/_Caron For the full text of finalist essays and for information about applying for next year’s awards, see Science Online at http://scim.ag/eppendorf.
References 1. C. Mora, D. P. Tittensor, S. Adl, A. G. B. Simpson, B. Worm, PLoS Biol. 9, e1001127 (2011). 2. J. O’Keefe, J. Dostrovsky, Brain Res. 34, 171 (1971). 3. M. Fyhn, S. Molden, M. P. Witter, E. I. Moser, M.-B. Moser, Science 305, 1258 (2004). 4. T. Hafting, M. Fyhn, S. Molden, M. B. Moser, E. I. Moser, Nature 436, 801 (2005). 5. E. I. Moser, E. Kropff, M.-B. Moser, Annu. Rev. Neurosci. 31, 69 (2008). 6. L. M. Giocomo et al., Neuron 71, 589 (2011). 7. N. Ulanovsky, C. F. Moss, Nat. Neurosci. 10, 224 (2007). 8. M. M. Yartsev, M. P. Witter, N. Ulanovsky, Nature 479, 103 (2011).
1 NOVEMBER 2013
VOL 342
SCIENCE
Published by AAAS
9. F. Sargolini et al., Science 312, 758 (2006). 10. T. Solstad, C. N. Boccara, E. Kropff, M.-B. Moser, E. I. Moser, Science 322, 1865 (2008). 11. G. Buzsáki, F. L. D. Silva, Prog. Neurobiol. 98, 241 (2012). 12. G. Buzsáki, X.-J. Wang, Annu. Rev. Neurosci. 35, 203 (2012). 13. P. Andersen, R. Morris, D. G. Amaral, T. Bliss, J. O’Keefe, The Hippocampus Book (Oxford Univ. Press, New York, 2007). 14. M. M. Yartsev, N. Ulanovsky, Science 340, 367 (2013).
www.sciencemag.org
10.1126/science.1245809
CREDITS: (TOP) ARTHUR COHEN PHOTOGRAPHY; (MIDDLE) COURTESY OF DANIEL BENDOR; (BOTTOM) GEORDIE WOOD
continuous [see the figure (A), middle], and the firing patterns of bat grid cells were not theta-modulated [see the figure (A), right], which put them in striking contrast to prerequisites of the oscillatory-interference models. Thus, our study in bats (8) allowed for causal examination of a major class of models that were based solely on data from rats. After establishing the existence of place cells and grid cells in a two-dimensional (2D) environment, we wanted to go one step further and ask: How is the complete 3D volumetric space represented in the mammalian hippocampal formation? This question is pivotal because many animals on our planet, whether in air, in water, or on land, move in 3D environments. However, all studies conducted to date were in either 1D or 2D environments (13), which left this question unresolved. The bat’s flight capability provided us with a unique opportunity to address this question. We focused on the hippocampus and developed the technology to record the activity of single place cells in freely flying bats (14) [see the figure (B), left]. We found that individual place cells provided a stable and nearly isotropic representation of the animal’s position in 3D space. Each place cell fired mainly in a single restricted region of the 3D environment, and all axes were represented with similar resolution [see the figure (B), middle]. Furthermore, each place cell activated in a different location, and the combined activity of multiple place cells represented the 3D environment uniformly [see the figure (B), right]. We further found that the firing patterns of 3D place cells were not theta-rhythmic during flight, which supported our previous f indings from the crawling bats and strongly argued against the cross-mammalian generality of oscillatory-based temporal codes for spatial representation (14). The importance of our findings is fourfold. First, they support the generality of the place-cell and grid-cell phenomena across mammals. Second, they argue in favor of a rate-coding mechanism underlying the formation of spatial firing patterns in these cell types and, by this, constrain the possible computations responsible for their generation. Third, they reveal a coding mechanism for 3D space in the mammalian brain. Finally, they demonstrate that the use of novel animal models in neuroscience can complement existing knowledge and provide insights into the inner workings of the brain.
Space Bats: Multidimensional Spatial Representation in the Bat
A novel animal model, the bat, is used to elucidate place-cell and grid-cell phenomena.
Michael M. Yartsev
Magnitude
(Right) Power-spectrum of a single grid-cell spike-train autocorrelogram showing no power in the theta frequency range (5 to 11 Hz, highlighted in red). (B) (Left) Illustration of a bat flying with the custom-made telemetry system, drawn to scale Princeton Neuroscience Institute, Princ- [Illustration: S. Kaufman]. (Middle) Color-coded three-dimensional firing-rate map, with peak rate indicated, of a single eton University, Princeton, NJ 08540, USA. 3D place cell recorded from the hippocampus of a freely flying bat. (Right) All place fields (individual colored convex hulls) E-mail: [email protected] recorded from the hippocampus of a single bat. Figures in (A) and (B) are adopted from (8) and (14), respectively.
www.sciencemag.org
SCIENCE
VOL 342
Published by AAAS
1 NOVEMBER 2013
573
Downloaded from www.sciencemag.org on March 17, 2015
I
t is estimated that more than 8 million tinuous theta oscillations (7, 8), perhaps different species reside on our planet, grid cells might also exist in their absence. many of which live a very different lifeThis would causally argue against the validstyle (1). But whether on the ground, in the ity of the oscillatory-interference class of ocean depths, or in the sky, all animals must models or, at least, against their generality have knowledge of their whereabouts to suracross mammals. We recorded the activvive. The possible mechanisms subserving ity of single MEC neurons in bats crawling this core function and how these are imple- Eppendorf and Science are pleased to present the inside a large arena (8) and found many of mented in the brain have been long-standing prize-winning essay by Michael M. Yartsev, the 2013 them to be grid cells [see the figure (A), left] questions in neuroscience. winner of the Eppendorf and Science Prize for Neu- with properties strikingly similar to those Scientists began studying the neural basis robiology. previously described in rats. We further of spatial representation with the report of found in the bat MEC all the other spatial neurons in the rat dorsal hippocampus that to the grid formation. Two major classes cell types previously described in the rat fired when the animal entered a specific loca- of computational models were proposed to MEC, such as neurons that encode the anition (2). These neurons were called “place account for this phenomena (6): Attractor- mal’s head orientation (9) and the borders cells.” More recently, scientists discovered based network models and single-neuron, of its current environment (10). We even “grid-cells” in the rat medial entorhinal “oscillatory-interference” models relying found many of the same neural oscillations cortex (MEC), which activate in multiple on theta-band (5 to 11 Hz) oscillations. The previously reported in the rat, such as highlocations, all arranged on the vertexes of a latter class of models received much exper- frequency “ripple” oscillations (11) and fast hexagonal grid (3, 4). Place cells and grid imental attention in rodent studies, but all and slow gamma oscillations (12). However, cells are widely considered key elements of evidences were of a correlative nature (6). as we had hypothesized, theta oscillations the mammalian spatial representation sys- We reasoned that, because place cells in bat in the bat were very different from those in tem (5), yet their detailed properties have hippocampus exist in the absence of con- rats. Theta oscillations in the bat were not been studied almost exclusively in rodents. This convergence in 1.6 Hz Theta choice of an animal model has A occluded the understanding of which neuronal mechanisms involved in spatial representation generalize across species and whether different solutions have been reached by different brains. To explore this ques0 10 20 30 tion and to study the functional Frequency (Hz) properties of these cell types, we B 15 Hz used a novel mammalian animal model, the bat. The findings in the bat helped revise our understanding of this important circuitry and show how divergence can powerfully complement convergence in the choice of animal models in neuroscience. We first asked whether the Properties of spatial cell types in crawling and flying bats. (A) (Left) Color-coded two-dimensional firing-rate map, with bat can provide insight into the peak rate indicated, of a single grid cell recorded from the MEC of a crawling bat. (Middle) Raw local field potential trace recorded from the MEC of a crawling bat showing a single intermittent theta-bout (top: 20-s trace; bottom: 7-s zoom-in). neural mechanisms giving rise
ESSAY
574
2013 Grand Prize Winner The author of the prize-winning essay, Michael Yartsev, received his undergraduate and master’s degrees in biomedical engineering from Ben-Gurion University in 2007. For his Ph.D., he joined the lab of Dr. Nachum Ulanovsky at the Weizmann Institute of Science. There, he recorded the activity of single neurons from the hippocampal formation of freely behaving and flying bats to study the underlying neural mechanisms of spatial memory and navigation in the mammalian brain. Since 2012, Dr. Yartsev is a C. V. Starr Fellow at the Princeton Neuroscience Institute at Princeton University where he is conducting postdoctoral work in the lab of Prof. Carlos Brody studying the neural basis of decision-making.
Finalists Daniel Bendor for his essay, “Play it again, brain.” Dr. Bendor is a lecturer in the Department of Cognitive, Perceptual, and Brain Sciences and the Institute of Behavioral Neuroscience at University College London. Dr. Bendor received his Ph.D. from Johns Hopkins University under the mentorship of Dr. Xiaoqin Wang, studying temporal processing in auditory cortex and the neural correlate of pitch and flutter perception. For his postdoctoral research, he investigated the role of the hippocampus in memory encoding and consolidation, while working with Dr. Matthew Wilson at the Massachusetts Institute of Technology. He has recently started his own lab at University College London, where his research focuses on how neural ensembles encode perceptual and memory-related information. http://scim.ag/_Bendor Sophie Caron for her essay, “Brains don’t play dice—or do they?” Dr. Caron is currently a postdoctoral fellow in the Department of Neuroscience at Columbia University. Sophie grew up in St-Blaise-sur-Richelieu in Canada and earned a B.Sc. in Biochemistry at the Université de Montréal. She moved to New York City to study the developmental mechanisms behind the diversification of sensory neurons in the laboratory of Dr. Alexander Schier at New York University and, later, Harvard. Having completed her Ph.D., Sophie joined the laboratory of Dr. Richard Axel at Columbia University, where she studies how the information gathered through the senses is represented in higher brain centers; in particular, those involved in memory. http://scim.ag/_Caron For the full text of finalist essays and for information about applying for next year’s awards, see Science Online at http://scim.ag/eppendorf.
References 1. C. Mora, D. P. Tittensor, S. Adl, A. G. B. Simpson, B. Worm, PLoS Biol. 9, e1001127 (2011). 2. J. O’Keefe, J. Dostrovsky, Brain Res. 34, 171 (1971). 3. M. Fyhn, S. Molden, M. P. Witter, E. I. Moser, M.-B. Moser, Science 305, 1258 (2004). 4. T. Hafting, M. Fyhn, S. Molden, M. B. Moser, E. I. Moser, Nature 436, 801 (2005). 5. E. I. Moser, E. Kropff, M.-B. Moser, Annu. Rev. Neurosci. 31, 69 (2008). 6. L. M. Giocomo et al., Neuron 71, 589 (2011). 7. N. Ulanovsky, C. F. Moss, Nat. Neurosci. 10, 224 (2007). 8. M. M. Yartsev, M. P. Witter, N. Ulanovsky, Nature 479, 103 (2011).
1 NOVEMBER 2013
VOL 342
SCIENCE
Published by AAAS
9. F. Sargolini et al., Science 312, 758 (2006). 10. T. Solstad, C. N. Boccara, E. Kropff, M.-B. Moser, E. I. Moser, Science 322, 1865 (2008). 11. G. Buzsáki, F. L. D. Silva, Prog. Neurobiol. 98, 241 (2012). 12. G. Buzsáki, X.-J. Wang, Annu. Rev. Neurosci. 35, 203 (2012). 13. P. Andersen, R. Morris, D. G. Amaral, T. Bliss, J. O’Keefe, The Hippocampus Book (Oxford Univ. Press, New York, 2007). 14. M. M. Yartsev, N. Ulanovsky, Science 340, 367 (2013).
www.sciencemag.org
10.1126/science.1245809
CREDITS: (TOP) ARTHUR COHEN PHOTOGRAPHY; (MIDDLE) COURTESY OF DANIEL BENDOR; (BOTTOM) GEORDIE WOOD
continuous [see the figure (A), middle], and the firing patterns of bat grid cells were not theta-modulated [see the figure (A), right], which put them in striking contrast to prerequisites of the oscillatory-interference models. Thus, our study in bats (8) allowed for causal examination of a major class of models that were based solely on data from rats. After establishing the existence of place cells and grid cells in a two-dimensional (2D) environment, we wanted to go one step further and ask: How is the complete 3D volumetric space represented in the mammalian hippocampal formation? This question is pivotal because many animals on our planet, whether in air, in water, or on land, move in 3D environments. However, all studies conducted to date were in either 1D or 2D environments (13), which left this question unresolved. The bat’s flight capability provided us with a unique opportunity to address this question. We focused on the hippocampus and developed the technology to record the activity of single place cells in freely flying bats (14) [see the figure (B), left]. We found that individual place cells provided a stable and nearly isotropic representation of the animal’s position in 3D space. Each place cell fired mainly in a single restricted region of the 3D environment, and all axes were represented with similar resolution [see the figure (B), middle]. Furthermore, each place cell activated in a different location, and the combined activity of multiple place cells represented the 3D environment uniformly [see the figure (B), right]. We further found that the firing patterns of 3D place cells were not theta-rhythmic during flight, which supported our previous f indings from the crawling bats and strongly argued against the cross-mammalian generality of oscillatory-based temporal codes for spatial representation (14). The importance of our findings is fourfold. First, they support the generality of the place-cell and grid-cell phenomena across mammals. Second, they argue in favor of a rate-coding mechanism underlying the formation of spatial firing patterns in these cell types and, by this, constrain the possible computations responsible for their generation. Third, they reveal a coding mechanism for 3D space in the mammalian brain. Finally, they demonstrate that the use of novel animal models in neuroscience can complement existing knowledge and provide insights into the inner workings of the brain.
