Subpopulations of neurons fire at specific geographical locations, providing a mental map of an animal's position in space. The finding that the circuitry can also support auditory maps sheds light on the neuronal structure of cognition. See Letter p.719
All animals face the challenge of navigating their environment, and are aided by an internal, mental map of where they have been and where they are going. Neurons in a brain region called the hippocampal formation are thought to form the substrate of such maps in mammals. But whether mental maps can also be used to organize other types of information is not known. On page 719, Aronov et al.1 provide evidence that the same neuronal circuitry can be used for both spatial and non-spatial mental maps in rats.
Neurons in the hippocampal formation, which includes the hippocampus and the adjacent entorhinal cortex2, provide an intricate representation of an animal's location in space. Hippocampal place cells and entorhinal grid cells fire at specific physical locations as a rat traverses an environment3,4. Some researchers think that these neurons function primarily as the brain's navigational system. However, others5,6 suggest that these neuronal ensembles might instead contribute in a more general way to the organization of information. This hypothesis accordingly raises the question of whether such neurons might represent non-spatial aspects of experience in a map-like way, similarly to their representation of space.
To address this question, Aronov et al. investigated whether neurons in the hippocampal formation can represent a 'map' of a non-spatial environment, choosing for their example an auditory landscape. They trained rats to use a joystick to manipulate a tone that got higher the longer the joystick was pressed. The rats were rewarded with a drop of water for releasing the joystick when the sound reached a particular frequency range. In this way, the authors could analyse the pattern of neuronal activity that occurred as the frequency changed, and compare this with the neuronal responses that represent a rat's changes in position as it navigates during a foraging task. To ensure that the rats were responding to a particular frequency and not simply to a set period of elapsed time, the researchers varied the speed at which the frequency increased across trials.
Just as neurons in the hippocampal formation respond when a rat is in a particular physical space, neurons fired in response to a particular tone in the auditory task (Fig. 1). Some of the tone-specific neurons in the auditory task were also place-specific neurons in the foraging task, suggesting that a common neuronal network supports both types of information map.
Next, Aronov and colleagues asked whether the neurons were merely responding to a particular frequency (as one might expect for simpler sensory areas of the brain), or whether the neuronal responses depended on a rat being engaged in a behavioural task. To test this, the authors presented rats that were not performing the joystick task with the same sound sweeps. Very few hippocampal cells were active during passive playback. Interestingly, when the investigators gave rats a water reward at the end of passive playback, hippocampal neurons did show frequency-modulated activity, albeit in fewer cells and with less precision than during the behavioural task. Thus, the predictability of reward might have directed the animals' attention to the stimuli. This attention might be a general prerequisite for forming a neuronal map.
The authors propose that a crucial feature of the mapping observed in their task is that frequencies are organized along a continuum. Just as the hippocampal map adopts the structure of the tonal scale, it might be expected that this network could map any stimulus dimension that has some graded, ordinal structure — such as light to dark, or hot to cold. However, the hypothesis that a continuum is important also places heavy constraints on the type of organization that the hippocampal formation can map. In particular, this hypothesis supposes that the hippocampus maps particular features of a given stimulus, rather than a temporally ordered series of discrete events. An alternative interpretation would be that the map tracks progress towards the reward, using the tones as milestones.
Future work could address this issue by changing the way in which tones progress as the rat presses the joystick — for instance, following an arcing pattern instead of merely increasing. If neuronal activity does truly represent a map of frequency space, each tone-specific neuron would fire twice because the same frequency would be played on either side of the arc. Alternatively, neuronal activity might track progression through the tonal series from the beginning of the trial to the reward. This distinction would shed light on whether hippocampal processes are constrained by the intrinsic features of their inputs (representing changes in frequency or space, for example), or whether this brain region can form relational maps that link arbitrary stimuli together.
Further questions are also inspired by the authors' data. To what extent can content-specific neuronal subpopulations, such as those that fire at certain times or frequencies, truly be considered to form a 'map'? What would be the purpose of such maps? Aronov et al. suggest that frequency representations have a key property that makes them map-like: the combined activity of the neuronal population completely reflects the progressing events of the task. Thus, at any point in time, a downstream brain region could decode the precise task phase by reading out which neurons are active within the population. This would imply that sequential neuronal activity represents the progressive flow of experience during the behavioural task, linking events into a singular memory — which may offer a more fundamental description of hippocampal activity than the activity patterns of individual neurons.
The current study provides strong evidence at the single-neuron level that activity in the hippocampal formation can reflect organization of non-spatial content. These findings add to a growing body of literature that indicates that the hippocampus can provide a structured organization for experience beyond navigation. Along with place cells that represent space, studies7,8 have identified hippocampal cells that are active at distinct time intervals during memory tasks. More recently, the hippocampal formation has been shown to be involved in humans' structured organization of social9 and conceptual relationships10.
Taken together, Aronov and colleagues' work expands our understanding of the function of the hippocampal formation. Future studies that extend this work to other species and identify the conditions that promote sequential firing within neuronal ensembles will help to clarify how these hippocampal representations organize information and support memory formation. Footnote 1
Aronov, D., Nevers, R. & Tank, D. W. Nature 543, 719–722 (2017).
Amaral, D. & Lavenex, P. in The Hippocampus Book (eds Andersen, P., Morris, R., Amaral, D., Bliss, T. & O'Keefe, J.) 37–114 (Oxford Univ. Press, 2006).
O'Keefe, J. & Dostrovsky, J. Brain Res. 34, 171–175 (1971).
Hafting, T., Fyhn, M., Molden, S., Moser, M.-B. & Moser, E. I. Nature 436, 801–806 (2005).
Schiller, D. et al. J. Neurosci. 35, 13904–13911 (2015).
Buffalo, E. A. Hippocampus 25, 713–718 (2015).
Pastalkova, E., Itskov, V., Amarasingham, A. & Buzsáki, G. Science 321, 1322–1327 (2008).
MacDonald, C. J., Carrow, S., Place, R. & Eichenbaum, H. J. Neurosci. 33, 14607–14616 (2013).
Tavares, R. M. et al. Neuron 87, 231–243 (2015).
Constantinescu, A. O., O'Reilly, J. X. & Behrens, T. E. J. Science 352, 1464–1468 (2016).
Related links in Nature Research
About this article
Cite this article
Rueckemann, J., Buffalo, E. Auditory landscape on the cognitive map. Nature 543, 631–632 (2017). https://doi.org/10.1038/543631a