View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Dispatch R841

Spatial Navigation: Head Direction swimming from several starting points to a platform hidden in a constant Cells Are Anchored by Gravity location. The slow learning and limited performance in the clinging task contrasts markedly with the Thalamic neurons that signal an animal’s direction of heading are crucial for rapid learning and highly flexible spatial navigation. Both directional coding and flexible use of spatial memory performance in the water maze, are upended, however, when a rat has to find its way while upside down. where rats escape directly from four starting points after only a few trials. Matthew Shapiro cells have been recorded in several The tenfold difference in learning rates interconnected structures that receive suggests that different mechanisms Adaptive behavior depends on vestibular inputs, including mammilary guide learning in the two navigation predicting accurate relationships nuclei, the anterior and lateral dorsal tasks. between percepts, actions, and nuclei of the , and the Spatial behavior can be guided by outcomes. Some of these relationships retrosplenial and entorhinal cortices. several mechanisms, including taxons generalize across niches, and neural Head direction cells fire at relatively and cognitive mapping [5]. Taxons refer mechanisms for computing predictions low base rates until the rat’s head is to directed responses with respect to a among them are conserved. pointing within about 60 of the cell’s single stimulus, such as approaching a Oculomotor systems predict how preferred angle, when the firing rate poster on a wall. Such cue approach retinal images change with eye or can increase to 60 spikes per second. strategies are impaired by lesions of head movements, and violating these The preferred heading angle of a given the dorsolateral that do not predictions impairs cognition in head direction cell is acquired as impair cognitive mapping [6]. surprising ways. I remember wearing animals explore an environment, is Navigation by cognitive mapping prism lenses that reversed expected stable in a constant environment, and combines spatial representations, and actual horizontal visual movement changes if prominent visual cues are defined by relationships among several and being utterly disoriented when I altered. stimuli, with signals tried to walk across a room. Analogous Heading signals are crucial for that indicate the direction and distance systems use spatial computations to ‘path integration’, the ability to track of movement. Rats demonstrate guide animals as they navigate about changing location by combining the cognitive mapping by finding familiar the environment, combining distance, distance and angles of self-generated spatial goals despite the removal of heading, and location signals to movement. Vestibular signals allow one or more visual cues, or after being track movement across the surface head direction cell signals to persist in placed in an unfamiliar start location of the world. A paper published the dark, but they ‘drift’ over time until a [5]. Both types of behavioral flexibility recently in Current Biology by Gibson familiar landmark is encountered that are typical in water maze tests, et al. [1] reports evidence that these corrects or resets them. Head direction suggesting that rats use relationships two-dimensional navigation systems cell activity is weakened or abolished among distal spatial cues to navigate are tethered by gravity. by lesions of the vestibular apparatus, to the hidden platform. Biological mechanisms for orienting especially the semi-circular canals; Gibson et al. [1] assessed the in gravity begin in the otolithic organs inactivation of the otoliths has less strategy used by the inverted rats by of the inner ear, where hair cells effect, which could imply that gravity placing the animals in an unfamiliar transduce linear acceleration. The plays a relatively unimportant role starting point at the center of the vestibular nuclei relay angular velocity in heading signals. The new work apparatus [1]. The rats headed in signals to other brainstem, thalamic, of Gibson et al. [1] overturns that random directions and showed no and cerebellar nuclei that convey implication, however, by showing that sign of learning after several trials, acceleration information to the spinal both head direction cell activity in the consistent with their inability to learn cord and the cerebral cortex. The anterodorsal thalamic nucleus and to escape from more than two starting dorsal tegmental and lateral navigation are severely disrupted when points. The inability to generalize in a mammillary nuclei integrate vestibular animals walk upside down. highly familiar environment, together with other perceptual and motor Gibson et al. [1] trained rats to cling with slow learning in the familiar start signals across levels of processing. to a circular wire mesh ‘ceiling’ and, task, suggest that the rats did not use Gravity, equivalent to a constant while upside-down, walk from starting a spatial mapping strategy, but rather acceleration, activates a subset of points at the edge of the apparatus to used a taxon strategy to approach these hair cells even when the head one of four potential escape hatches specific distal cues. Indeed, when a is still [2]. near the middle of each quadrant. The curtain blocked the distal cues from Head direction cells integrate correct escape hatch let the rat climb view, the upside down rats made vestibular signals with other percepts above the mesh and stand up. After random headings even from the to compute horizontal orientation. more than 100 training trials, the rats familiar starting points, as though Head direction cells act like neuronal learned to move directly to the escape the rats no longer had a sense of Geiger counters with receptive fields hatch from each of two different direction. Indeed, head direction that span whole environments and starting points, but they could not learn signals were lost when the rats were have one trigger feature — the animal’s four. The task is operationally similar upside down. heading angle. First discovered in the to the Morris water maze [4], in which Previous work showed that head rat presubiculum [3], head direction rats learn to escape from the water by direction cell coding was maintained Current Biology Vol 23 No 18 R842 when rats walked along a horizontal place fields [10], and hippocampal problems learning new facts and maze arm and then climbed up a lesions reduce the stability of head remembering recent events [18]. vertical wall [7], suggesting that direction cells [14]. Cognitive mapping theory proposes heading signals might be independent Computational models of navigation that spatial computations define the of gravity. Gibson et al. [1] recorded propose that recurrent activation of fundamental operations of memory anterodorsal thalamic nucleus cells as place, distance, and heading encodes systems, and that these other types rats explored a cylinder on the floor and retrieves spatial paths [5,13,15,16]. of memory rely on these. Relational of the testing room before and after The models suggest that moving memory theory proposes that spatial they performed the inverted through an environment activates memory exemplifies a general memory navigation task [1]. They found that interconnected head direction, grid, system that associates stimuli that the anterodorsal thalamic nucleus cells and place units so that specific overlap in time into events and had sharply-tuned directional firing trajectories are represented by sequential episodes [19]. The correlates when the rats walked on the direction–distance–place code computational differences between floor, but the signals were disrupted sequences and stored by synaptic spatial navigation and other forms when the same cells were recorded plasticity. Familiar trajectories could be of flexible memory remain unclear. when rats were upside down. The retrieved as place-direction-distance Consider the example of rats trained animals performed the inverted task sequences; for example, when placed to distinguish between two olfactory reliably and followed consistent in a familiar start location in a water sequences that begin and end trajectories from the two familiar maze, a rat’s active hippocampal with distinct odors but contain an starting points, but the peak firing rates cells would signal that location and overlapping set of smells in the of the head direction cells declined, activate head direction cells predicting middle [19]. background firing increased, and the platform direction. The head Compare that to another example directional tuning curves were noisy direction cells would activate grid in which rats are trained to distinguish and inconsistent. During the center cells that signal distance, and together between two spatial paths that begin probe tests, the directional tuning these would activate the next and end in different places but have an declined further. Head direction cell hippocampal spatial representation on overlapping middle section. Both tasks activity returned to normal when the the way to the platform, and so on. are impaired by hippocampal damage, animals walked upright in the cylinder Heading angles thereby predict future but to what extent do they require the afterwards, and the strong tuning locations by activating spatial same computations? If non-spatial curves were the same before and after sequences. memory processing depends upon the inverted tests. Despite their ability These mechanisms explain how navigation mechanisms, then any to use distal cues to guide navigation is accomplished by path manipulation that impairs navigation goal-directed behavior, the rats lost integration despite cue removal, should also impair nonspatial memory. head direction signals and cognitive when both head direction cell and Head direction cells should be crucial mapping when journeys started upside place signals are maintained, and why for the spatial task, as heading (along down. navigation is impaired when heading with goal) information is crucial for Navigation requires several signals are disrupted. Furthermore, linking successive places along each interconnected limbic structures [4,8]. the models predict that place fields, journey. If nonspatial memory Head direction cells interact with like head direction cells, should be processes depend upon navigation cells in the that signal disrupted in inverted rats, which has mechanisms, then head direction cells location, and medial not yet been tested. Head direction should be required for disambiguating cells that code distance [9,10]. Rats cell dysfunction could untether the olfactory sequences. Perhaps different with head direction cell dysfunction orientation of otherwise intact place odor sequences trigger different head do not learn the location of a hidden field maps, or disrupt activation of direction cells, and ‘navigation’ platform in the water maze, but readily place field sequences [16]. Computing proceeds through an olfactory space. learn to approach a visible platform novel routes requires associating goals If, however, head direction signals [11]. Hippocampal dysfunction causes with locations and headings, so would comprise one type of predictive, similar deficits [4]. Hippocampal presumably depend upon head directed association among others, neurons code locations via place direction signals [17]. then head direction cells should be fields, small regions in an environment That head direction cells and irrelevant to the olfactory sequence that elicit high firing rates when navigation depend on gravity task. visited by a rat [5]; 60 hippocampal has important implications for From this view, non-spatial relational cells can predict the location of a understanding the flexible use of memory tasks engage parallel inputs rat’s head within 2 cm2 [12]. Like memory in general. In rats and other to the hippocampus, so that disrupting head direction cells, place fields are species, the hippocampus, entorhinal spatial navigation selectively, by influenced by both external stimuli cortex, and dorsal thalamus are disrupting head direction cells, should and path integration [13]. The required for memory tasks that do not leave intact other forms of relational neuropsychological parallels suggest entail spatial navigation. Contextual memory. The study by Gibson et al. [1] that place and head direction signals memory retrieval of olfactory and provides a nice opportunity to interact, and indeed head direction visual associations, odor sequences, distinguish these possibilities, simply cells contribute to stable place fields transitive inference, and trace eyelid by testing the rats in an explicitly and vice versa. Anterodorsal thalamic conditioning are impaired by similar nonspatial memory task while they are nucleus lesions decrease the lesions in rats. People with inverted. If relational memory rides specificity and stability of hippocampal homologous brain damage have atop spatial navigation mechanisms, Dispatch R843 then it should fall when rats are upside 7. Stackman, R.W., Tullman, M.L., and Taube, J.S. 14. Golob, E.J., and Taube, J.S. (1999). Head (2000). Maintenance of rat head direction direction cells in rats with hippocampal or down and head direction cell signals cell firing during locomotion in the overlying neocortical lesions: evidence for are lost. If navigation exemplifies a vertical plane. J. Neurophysiol. 83, impaired angular path integration. J. Neurosci. broader set of memory computations, 393–405. 19, 7198–7211. 8. Clark, B.J., and Taube, J.S. (2009). Deficits in 15. Hasselmo, M.E., Giocomo, L.M., Brandon, M.P., then the loss of head direction cell landmark navigation and path integration after and Yoshida, M. (2009). Cellular dynamical signals should spare non-spatial lesions of the interpeduncular nucleus. Behav. mechanisms for encoding the time and place of Neurosci. 123, 490–503. events along spatiotemporal trajectories in relational memory. 9. Sargolini, F., Fyhn, M., Hafting, T., episodic memory. Behav. Brain Res. 215, McNaughton, B.L., Witter, M.P., Moser, M.B., 261–274. References and Moser, E.I. (2006). Conjunctive 16. Buzsaki, G., and Moser, E.I. (2013). Memory, 1. Gibson, B., Butler, W.N., and Taube, J.S. (2013). representation of position, direction, and navigation and theta rhythm in the The head-direction signal is critical for velocity in entorhinal cortex. Science 312, hippocampal-entorhinal system. Nat. Neurosci. navigation requiring cognitive map but not for 758–762. 16, 130–138. learning a spatial habit. Curr. Biol. 23, 10. Calton, J.L., Stackman, R.W., Goodridge, J.P., 17. Johnson, A., and Redish, A.D. (2007). Neural 1536–1540. Archey, W.B., Dudchenko, P.A., and ensembles in CA3 transiently encode paths 2. Cullen, K., and Sadeghi, S. (2008). Vestibular Taube, J.S. (2003). Hippocampal forward of the animal at a decision point. systems. Scholarpedia (http://www. instability after lesions of the head direction J. Neurosci. 27, 12176–12189. scholarpedia.org/article/Vestibular_system). cell network. J. Neurosci. 23, 18. Eichenbaum, H.B., and Cohen, N.J. (2001). 3. Ranck, J.B. (1984). Head-direction cells in 9719–9731. From Conditioning to Conscious Recollection the deep cell layers of dorsal presubiculum 11. Clark, B.J., Rice, J.P., Akers, K.G., (New York: Oxford Psychology Series - Oxford in freely moving rats. Soc. Neurosci. Abstr. Candelaria-Cook, F.T., Taube, J.S., and University Press). 10, 599. Hamilton, D.A. (2013). Lesions of the dorsal 19. Eichenbaum, H. (2013). Memory on time. 4. Morris, R.G., Garrud, P., Rawlins, J.N., and tegmental nuclei disrupt control of navigation Trends Cogn. Sci. 17, 81–88. O’Keefe, J. (1982). Place navigation impaired in by distal landmarks in cued, directional, and rats with hippocampal lesions. Nature 297, place variants of the Morris water task. Behav. 681–683. Neurosci. 127, 566–581. Department of Neuroscience, Mount Sinai 5. O’Keefe, J., and Nadel, L. (1978). The 12. Wilson, M.A., and McNaughton, B.L. (1993). Hippocampus as a Cognitive Map (Oxford: Dynamics of the hippocampal ensemble School of Medicine, One Gustave L. Levy Oxford U.P.). code for space. Science 261, 1055–1058. Place, New York, NY 10029, USA. 6. White, N.M. (2008). Multiple memory systems in 13. McNaughton, B.L., Battaglia, F.P., Jensen, O., E-mail: [email protected] the brain: cooperation and competition. In Moser, E.I., and Moser, M.B. (2006). Path Learning and Memory: A Comprehensive integration and the neural basis of the Reference, J.H. Byrne, ed. (Oxford: Academic ‘cognitive map’. Nat. Rev. Neurosci. 7, Press), pp. 9–46. 663–678. http://dx.doi.org/10.1016/j.cub.2013.08.002

Visual Working Memory: Now You (Figure 1B). When the distracting landmark and the original target are See It, Now You Don’t then removed, analogous to walking down into the ravine, a fly will re-orient toward the original, but now absent Flies can form a visually-guided working memory. A new study shows that the target. Flies will go back to the original gene termed ellipsoid body open influences multiple signals to regulate a target if the distractor is present for less competence factor in the ellipsoid body to support normal working memory. than four seconds, suggesting that a seconds-long working memory allows Lily Kahsai and Troy Zars Here is the first sleight: tricking a fly to re-orient toward a disappeared flies into showing that they have a landmark (Figure 1C). It’s an age-old problem, trying to visually-guided working memory. How does a fly form this visual navigate through rough terrain with Evidence for this type of memory working memory? A first clue to the intermittent landmarks. You pick a can be seen in individual flies in the neural mechanism for this type of target, say a tall tree, and walk toward so-called disappearing landmark memory came from a mutant fly type it, only to have the tree disappear as paradigm. In this test, a single fly is that had a grossly misformed part of you move down into a ravine. There is put in a circular arena, about the size the brain. A mutation that alters the an idea of the right direction, and with of an end table, which is lined with LED structure of the ellipsoid body, called some level of error, you can predict lights controlled by a computer [2].If ellipsoid body open (ebo), has provided pretty well the proper track back to the arena is uniformly lit, the fly will ideas about multiple behaviors, the target. Humans can do this. walk around in random directions [2,3]. including premotor behaviors (for Impressively, some seemingly simple If, however, two vertical dark bars are example [5,6]). Among the abnormal animals can also use a working placed at 180 degrees from each other, behaviors of the mutant flies is a clear memory to re-orient toward a lost then the fly starts to walk back and defect in visually guided working target. A new study by Roland Strauss forth between the two landmarks. A memory [2]. What is surprising, and colleagues [1] reported in this issue fly will walk between the landmarks however, is that although the ebo gene of Current Biology demonstrates that in this modification of the Buridan is acting in the ellipsoid body (more the fruit fly Drosophila melanogaster Paradigm for hours, approaching first on this brain structure next), its critical can use visual landmarks to establish one landmark then turning around and role in working memory is independent a seconds-long working memory and going to the other [2,4] (Figure 1A). of the structural change in this brain elucidates a novel cellular and neural Now, if a distracting landmark structure seen in the mutant flies [1].It circuit mechanism to support this type appears on the surface of the arena, turns out that the structural change in of memory. a fly will orient toward the new stripe the ellipsoid body of the ebo mutant