← all systems · see it in The System
Path integration
Path integration is the method thought to be used by animals for dead reckoning: integrating all changes of direction with the distance covered on a journey into a continually updated estimate of position relative to a starting point (the home vector), enabling a straight-line return.
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | kind: not dated in the sources read; the insect central-complex outline is over 350 million years old |
| born precision | lower bound only, and for the circuit, not the behaviour |
| boundary | Inside: the running computation that turns self-motion and a direction reference into a position estimate relative to a start (the home vector), its working memory, and its readout for steering, search, dance and report. Outside: the senses that feed it (eyes, legs, vestibular organs), the animal that carries it, the landmark and view memories it works beside, and the world it is run in. |
| nested level | A navigation strategy inside an animal's nervous system: built from a heading code, a speed or distance code, an accumulator and a steering readout; carried by insect central complexes and mammalian head-direction, grid and place cell systems; part of each animal's navigation alongside view-based guidance and search. |
| scale | organism |
Measures (24) published science
| path-integration memory lifetime | about 24 h, declining rapidly thereafter |
| view memory lifetime (desert ants) | the entire lifetime |
| age of the central-complex circuit outline | more than 350 million years |
| central-complex columns | 16-18 vertical columns |
| E-PG heading cells | eight per hemisphere columnar neurons (Drosophila) |
| TN speed cells | four neurons per brain (bee) |
| CPU4 cells per column | a minimum of 18 cells (examined bee species) |
| positional uncertainty growth (Cataglyphis) | >310 cm² per 1 cm walked (95% CI) |
| search width W vs nest recognition (Cataglyphis) | 3.8 m vs 2-3 cm m / cm |
| optimal search radius | proportional to the fourth root of total search time scaling |
| maximum search radius (Cataglyphis) | proportional to t^0.5 scaling |
| ant search-start distances compared | 14 vs 28 m |
| accurate integration range (grid attractor model) | about 10-100 m and 1-10 min m, min |
| periodic grid model error | <15 cm over about 260 m and 20 min (grid period about 48 cm) |
| grid tuning in the dark | relatively accurate over about 100 m and 10 min m, min |
| rat homing without cues | after random foraging trajectories 1-3 m long |
| blindfolded walking circle diameter | < 20 m |
| idiothetic reach limit (walking humans) | approximately 100 |
| AntBot mean homing error | 0.67 % of trajectory |
| Stangl participants | 30 young (19-26 y) and 26 older (62-78 y) people |
| path-integration test vs cognitive tests for Alzheimer's biomarker status | AUC 0.90 vs 0.57 area under the curve |
| AntBot compass | 2 UV pixels with rotating polarised filters, equivalent to 374 photosensors |
| Megalopta ground speed | 20 cm/s |
| Megalopta bees recorded | 160 bees |
Open contentions (7) open contention
Where sources or readings disagree. Both stay on the map.
- C-1691 Can all insects keep track of the way home, or only homing ones?
booked: The map leaves it open: homing insects only, or a basic ability of every insect brain.
the other reading: homing insects only; all insects, a basic ability - C-1693 Do desert ants add up their path exactly or approximately?
booked: The map keeps both: a simple approximation with small errors, and true vector computation in the fly.
the other reading: approximation; vector computation - C-1714 Do the brain's grid cells keep track of the way home?
booked: The map books them as the mammalian candidate and leaves open whether they compute it.
the other reading: the encyclopedia: 'Whether place and grid cells actually compute a path integration signal remains to be seen'; Gil et al. 2018: disrupted grid firing accounts for impaired path integration in mice; Burak and Fiete 2009: earlier models' errors 'cast doubt'; their attractor networks integrate accurately over 10-100 m - C-1721 Do animals use Earth's magnetic field to keep track of the way home?
booked: The map keeps the mole-rat evidence and leaves the insect case open.
the other reading: used (mole rat); unclear (insect PI) - C-1724 Where does the error in path integration come from?
booked: The map keeps both: noise in the movement signals (people) and noise inside the update (ants, inferred).
the other reading: input noise (humans); internal update noise (ants, inferred) - C-1725 Is path integration a separate module in the brain?
booked: The map leaves it open: a separate module, or interwoven with search, dance and landmark memories.
the other reading: isolated module; interwoven - C-1726 How do animals weigh their path sense against what they see?
booked: The map leaves it open between weighting by certainty of position and by certainty of direction.
the other reading: positional certainty weighting; directional certainty weighting
Valence: what it trades, and with whom (12) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Insect brain via sys:electrochemical-gradient · Path integration gain
- Entorhinal cortex via sys:electrochemical-gradient · Path integration gain
- planetary magnetic field (geodynamo) via sys:planetary-magnetic-field · Path integration gain
- Sun via sys:electromagnetic-field · Path integration gain
- Human (Homo sapiens) via sys:sensory-environment · Path integration gain
- Hippocampus (memory) via sys:electrochemical-gradient · Path integration gain
- Honeybee (western honey bee, Apis mellifera) via sys:mechanical-contact · Path integration gain
- Desert ant (Cataglyphis) via sys:sensory-environment · Path integration gain
- Alzheimer's disease via sys:sensory-environment · Path integration loss
- Robot (learning robot) via sys:electromagnetic-field · Path integration even
- Central complex (insect brain) via sys:electrochemical-gradient · Path integration gain
- View-based navigation (landmark guidance) via sys:sensory-environment · Path integration mixed
Partners and parts published science
Built from / builds (7)
- Grid cell
- Head direction cell
- Central complex (insect brain)
- Odometer (step integrator or optic-flow speed cells)
- Systematic search
- Path integration and view-based guidance compete for the steering output
- Landmarks and self-motion compete to set the grid map
Sits inside / holds (6)
- Human (Homo sapiens)
- Honeybee (western honey bee, Apis mellifera)
- Fruit fly (Drosophila melanogaster)
- Desert ant (Cataglyphis)
- Animal navigation
- Blind mole rat
expressed by (9)
- Scaffolding/Enablement (Meta-Tooling)
- Interoperability/Transduction (Bridging)
- Sensing/Scanning
- Search/Foraging
- Mapping/Modeling
- Navigation/Pathfinding
- Multi-Scale Data Fusion / Triangulation
- Redundancy & Diversity
- Quality Control & Error Correction
floor filled by (15)
- Gravitational field
- planetary magnetic field (geodynamo)
- Sun
- Thalamus
- Brainstem (midbrain, pons, medulla)
- Grid cell
- Hippocampus (memory)
- Muscle spindle
- Compound eye
- Head direction cell
- Vestibular system
- Central complex (insect brain)
- Systematic search
- View-based navigation (landmark guidance)
- Sky-compass pathway
Kinds (9)
- Desert-ant path integration
- Allothetic path integration
- Idiothetic path integration
- Honeybee path integration (personal and communal vectors)
- Mammalian path integration (head-direction, grid and place cells)
- Human path integration
- Egocentric coordinate system
- Allocentric static vectorial representation
- Dead reckoning (sailors)
Notes published science
- A desert ant's sense of where home is gets fuzzier at a startling rate: by the search data, its uncertainty grows by more than 310 cm², about half an A4 page, for every centimetre it walks, far more than sensory noise can explain (src:heinze-2018).
- Lost people really do walk in circles, and not because one leg is longer: blindfolded walkers looped in circles under 20 m across, in no fixed direction, and in forest and desert people circled only when they could not see the sun (src:souman-2009).
- Your path integrator may tell on your brain: an immersive path-integration test separated people with mild cognitive impairment who carry Alzheimer's biomarkers from those who don't with an AUC of 0.90, against 0.57 for the best standard cognitive test (src:howett-2019).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
path integration: from self-motion and compass to the way home (reader's view)
- L1: The world the animal moves through: the sky (sun and the polarisation pattern, fixed at infinite distance so it changes only when the animal turns), the ground under its legs and its slopes, the scene flowing past the eyes, gravity and inertial accelerations, Earth's magnetic field
- L2: The senses that read self-motion and direction: dorsal-rim ommatidia (polarisation), the ventral eye (optic flow), leg joints or the walking pattern generator (steps), the vestibular organs, muscle and joint receptors
- L3: Routing: the compass pathway from optic lobe to the central complex; speed neurons from the ventromedial and lateral protocerebrum to the noduli; in mammals brainstem nuclei and cortical structures
- L4: Self-motion estimates at their ports: an ordered E-vector array in the protocerebral bridge, TN speed cells in the noduli, PFN cells receiving locomotor velocity, the anterior dorsal thalamic nucleus for mammalian heading
- L5: Heading and world-centred velocity: a ring-attractor bump that encodes heading (E-PG cells; head-direction cells in mammals), and body-centred velocity rotated into world coordinates
- L6: The integrator: the running home vector, distance and direction stored together in direction-locked accumulators (CPU4 / PFN in insects), grid cells in mammals; working memory of about a day
- L7: The way home used: compare desired with current heading and steer (CPU1 to the lateral accessory lobe), weigh the vector against views by reliability, search when the vector runs out, dance the vector, look back at the hidden nest, or (people) say how far and turn to point
the boardroom (shared floors)
- L5: Heading: a ring-attractor bump encoding current heading, set by compass and landmarks and moved by self-rotation (E-PG and P-EN cells in the insect central complex; head-direction cells in mammals)
- L5: World-centred velocity: body-centred speed signals multiplied with heading and rotated into world coordinates (PFNd, PFNv, hΔB in flies; TN cells in bees; pooled self-motion estimates in humans)
- L6: The integrator: the home vector held as distance and direction together in direction-locked accumulators (CPU4 / PFN, a ring of static reference vectors), or in mammals a grid-cell map that persists without landmarks; possibly separate outbound and inbound accumulators
- L7: Use: compare the vector's desired heading with the current heading and steer; weigh it against views and stored vectors by reliability (the input with best signal to noise dominates); hand over to search when the vector is run off; in people, the conscious estimate reported
way in: Efference copy (what the motor system commanded)
- L1: Motor commands: what movements were commanded and executed
- L4: Locomotor brain regions hand velocity to the PFN cells of the fan-shaped body
way in: Landmarks, views and boundaries (the correcting channel)
- L1: The visual scene, external landmarks and the boundaries of the space
- L3: Ring (TL) neurons carrying the visual scene in flies, the compass in other insects
- L4: Onto the heading cells (E-PG); in mammals border cells as the error-correcting port
way in: Magnetic field (geomagnetic compass)
- L1: Earth's magnetic field, its horizontal component
way in: Memory (stored vectors and views)
- L1: Long-term memories: averaged vectors and nest-associated views (views last a lifetime in desert ants)
- L4: The mushroom body as the likely store; input neurons to the upper central body from areas receiving mushroom-body output
way in: Optic flow (the world sliding past the eye)
- L1: The image of the world moving across the eye as the animal travels
- L2: The ventral region of the eye (ants); the eyes of flying insects; human vision
- L3: Relay from the ventromedial and lateral protocerebrum to the noduli
- L4: TN1 and TN2 speed cells, four per brain, tuned 45° either side of the midline front and back
way in: Proprioception (muscles and joints)
- L1: The body's own limb positions and movements
- L2: Muscle and joint receptors
- L3: Brainstem nuclei
- L4: Cortical structures that estimate angular and linear movement velocity
way in: Sky compass (polarised skylight and sun)
- L1: The polarisation pattern of the blue sky (and the sun), at infinite distance
- L2: Dorsal rim ommatidia of the compound eye, tuned to the E-vector
- L3: The compass pathway: neurons indirectly linking the optic lobe to the lower central body
- L4: An ordered array of E-vector tunings across the protocerebral bridge columns
way in: Stride count (the step integrator)
- L1: The ground under the legs, including its slopes: the horizontal projection is what gets summed
- L2: Unknown which: proprioceptors in the body joints, or the walking pattern generator
way in: Vestibular (rotation and acceleration)
- L1: Movement in three dimensions: rotations and accelerations of the head (inertial cues)
- L2: The vestibular organs
- L3: Brainstem nuclei where self-motion cues are integrated
- L4: Anterior dorsal thalamic nucleus, critical for generating the directional signal
way out: Look-back gaze in learning walks
- L1: Turns and pirouettes that face the nest and take in nest-oriented views
- L7: The path integrator read out to aim the gaze at the hidden nest
way out: Report (people: say how far, turn to point)
- L1: A spoken distance and a body turned toward the start
way out: Steering (legs and wings)
- L1: The body turns and runs or flies toward home
- L5: Lateral accessory lobe premotor neurons (flip-flop activity in moths)
- L7: CPU1 output cells compare memory with current heading; a hemispheric imbalance starts the turn
way out: Systematic search
- L1: Ever-increasing loops that return near the start of the search
- L7: When the whole vector is run off without finding the goal, search begins
way out: Waggle dance (the vector told to nestmates)
- L1: Waggle runs on the comb: waggle duration proportional to distance, angle to vertical giving direction to the sun
- L7: The communal vector chosen for the dance, possibly apart from the personal vector
Sources (20) published science
Hardcastle K, Ganguli S, Giocomo LM (2015) Environmental boundaries as an error correction mechanism for grid cells. Neuron 86:827-839. doi:10.1016/j.neuron.2015.03.039 (behind its links)
Gil M, Ancau M, Schlesiger MI, et al. (2018) Impaired path integration in mice with disrupted grid cell firing. Nat Neurosci 21:81-91. doi:10.1038/s41593-017-0039-3 (behind its links)
Kim IS, Dickinson MH (2017) Idiothetic path integration in the fruit fly Drosophila melanogaster. Curr Biol 27:2227-2238. doi:10.1016/j.cub.2017.06.026 (behind its links)
Kimchi T, Etienne AS, Terkel J (2004) A subterranean mammal uses the magnetic compass for path integration. PNAS 101:1105-1109. doi:10.1073/pnas.0307560100 (behind its links)
Cheung A, Vickerstaff R (2010) Finding the way with a noisy brain. PLoS Comput Biol 6:e1000992. doi:10.1371/journal.pcbi.1000992 (behind its links)
Srinivasan MV, Zhang S, Altwein M, Tautz J (2000) Honeybee navigation: nature and calibration of the 'odometer'. Science 287:851-853. doi:10.1126/science.287.5454.851 (behind its links)
Burak Y, Fiete IR (2009) Accurate path integration in continuous attractor network models of grid cells. PLoS Comput Biol 5:e1000291. doi:10.1371/journal.pcbi.1000291 (behind its links)
Souman JL, Frissen I, Sreenivasa MN, Ernst MO (2009) Walking straight into circles. Curr Biol 19:1538-1542. doi:10.1016/j.cub.2009.07.053 (behind its links)
Lu J, Behbahani AH, Hamburg L, et al. (2022) Transforming representations of movement from body- to world-centric space. Nature 601:98-104. doi:10.1038/s41586-021-04191-x (behind its links)
Taube JS (1998) Head direction cells and the neurophysiological basis for a sense of direction. Prog Neurobiol 55:225-256. doi:10.1016/S0301-0082(98)00004-5 (behind its links)