Neural circuits for goal-directed navigation across species

Abstract

Across species, navigation is crucial for finding both resources and shelter. In vertebrates, the hippocampus supports memory-guided goal-directed navigation, whereas in arthropods the central complex supports similar functions. A growing literature is revealing similarities and differences in the organization and function of these brain regions. We review current knowledge about how each structure supports goal-directed navigation by building internal representations of the position or orientation of an animal in space, and of the location or direction of potential goals. We describe input pathways to each structure - medial and lateral entorhinal cortex in vertebrates, and columnar and tangential neurons in insects - that primarily encode spatial and non-spatial information, respectively. Finally, we highlight similarities and differences in spatial encoding across clades and suggest experimental approaches to compare coding principles and behavioral capabilities across species. Such a comparative approach can provide new insights into the neural basis of spatial navigation and neural computation.

Keywords: central complex; goal encoding; hippocampus; insect; place cells; rodents.

Figure 1:. Place cells of the hippocampal formation

(A) Schematic overview showing the mouse hippocampal place cell system. The hippocampus and its subdivisions (DG, CA1, CA2, and CA3) are indicated in a coronal cross section of the adult mouse brain. CA1 pyramidal neurons show spatially tuned activity characterized by an increase in firing rate as the mouse navigates across distinct locations in space (place fields). In the schematic on the right, the CA1 place cells are color coded as per their place field locations on the linear track in (B). (B) Task selective place maps emerge to orthogonally represent goal-oriented spatial navigation learning rules and contexts. In the illustrated odor cued goal oriented spatial navigation task, the mouse probes two distinct odor cues and learns to navigate on the same textured belt to distinct reward locations and operantly lick for sugar water rewards (task design and observations adapted from [26]). Distinct place cell ensembles are activated for the different odor trial types, showing task selective place map representations and remapping based on the reward context. Figure created with biorender.com.

Figure 2:. Information flow in the vertebrate navigation system

(A) Information flow in the vertebrate navigation system. (Left) Schematic of the mouse brain showing the topographical relationship between entorhinal cortex, subdivided into medial (blue) and lateral (orange) entorhinal cortices (MEC and LEC, respectively), and the hippocampus (green) in the temporal lobe. (Right) Horizontal cross-section showing direct and indirect information pathways. Direct inputs from the MEC and LEC convey multi-sensory spatial and context information to each of the hippocampal subfields. MEC and LEC inputs are anatomically segregated along the proximal-distal axis in CA1, but are integrated by the same neurons in CA3 and DG. The indirect tri-synaptic pathway routes EC information through the dentate gyrus (DG) via the mossy fiber inputs to CA3, and then from CA3 to CA1 via Schaffer collateral inputs. CA1 serves as the major output region of the hippocampus projecting mnemonic spatial and contextual ‘feedback’ information to several cortical areas, including the entorhinal cortices. (B) Functional separation between LEC and MEC showing distinct feature selective tuning. MEC neurons code for spatial features including grid position, Cartesian position, boundaries, head angle, and speed. LEC neurons respond robustly to objects, particularly novel objects and their displaced locations, salient cues such as odors or rewards (and punishments), and temporal structure of tasks. The hippocampal place cell system builds a map of space, with each cell showing selectivity to a particular spatial location (a place field). While grid cells have multiple firing fields, most CA1 neurons show only one. CA1 neurons also show selectivity to rewards, episodic sequences, and learning rules. Changes in context, such as the shape of a room where navigation occurs, lead to changes in place cell tuning. Such remapping results in reorganization at the ensemble level and the emergence of a new place map.

Figure 3:. Overview of the insect central complex

(A) Structure of the insect central complex. The ellipsoid body houses a heading representation that is broadcast to the protocerebral bridge , and then to the fan-shaped body. The fan-shaped body also received self-motion information vis the lateral accessory lobe and paired noduli (NO). Contextual information enters the fan-shaped body through a separate pathway. The fan-shaped body is thought to generate representations of goals in allocentric coordinates which is read out by neurons projecting back to the lateral accessory lobe to drive goal-directed steering. (B) Vector representations in the central complex. Neuropils of the central complex such as the fan-shaped body , shown here, are composed of ordered arrays of developmentally-related neurons that can represent vectors through calcium activity in their processes. The schematic depicts a class of hΔ type local neurons that arborize in two different columns of the fan-shaped body , with a separation of half the fan-shaped body (one neuron shown in dark bold to clarify its anatomical projections). Like all insect neurons, these have their cell bodies at the exterior of the brain and make arborizations within the structured neuropils of the central complex. Most such populations exhibit sinusoidal patterns of calcium activity (orange) restricted to rows of processes. The location of the peak of this sinusoid corresponds to a vector angle, while the intensity of the activity can correspond to a vector length (arrow at bottom). Ordered projection patterns across the rows of the fan-shaped body can allow for vector summation and rotation. Abbreviations: EB - ellipsoid body; FB - fan-shaped body; LAL - lateral accessory lobe; NO – paired noduli; PB - protocerebral bridge.

Figure 4:. Heading and steering systems of the central complex

(A) The compass system of the ellipsoid body. Compass neurons (also known as EPGs, top) arborize in tiles of the ellipsoid body. A bump of calcium activity across this array represents the heading of the fly relative to visual and mechanosensory landmarks (bottom) and rotates when the fly rotates. Inhibitory ring neurons (gray, top) carry sensory information about visual landmarks and wind direction to the compass system. Anti-hebbian plasticity between ring neurons and compass neurons that is gated by rotational movement allows the fly to learn new mappings between sensory cues and heading direction. Two copies of the ellipsoid body heading signal are transmitted to the protocerebral bridge, one to each side (see figure D). (B) Steering system of the central complex. PFL3 neurons (green) receive input in both the protocerebral bridge and fan-shaped body and output to descending neurons in the lateral accessory lobe that can drive left or right turns. The anatomy of these neurons allows them to compare the heading representation in the protocerebral bridge (purple) with a goal signal in the fan-shaped body (orange) and drive turning until these two are aligned. Abbreviations: EB - ellipsoid body; FB - fan-shaped body; LAL - lateral accessory lobe; PB - protocerebral bridge.

Figure 5:. Input pathways to the fan-shaped body

(A) Columnar input neurons of the fan-shaped body (PFNs) carry heading and self-motion information. Each neuron receives a heading signal in the protocerebral bridge and one or more sensory inputs in the paired noduli related to self-motion: optic flow, airflow, or proprioception. Fan-shaped body columnar neurons form a vector representation of self-motion that can be used to compute allocentric travelling direction within the fan-shaped body through vector summation. (B) Tangential neurons of the fan-shaped body carry predominantly non-spatial context cues such as odorants, tastants, and behavioral states. Activation of tangential neurons can alter navigation behavior presumeably by regulating the activity of different goal-encoding local neurons. Abbreviations: FB - fan-shaped body; NO – paired noduli; PB - protocerebral bridge.