Animal Cognition I

1
Raumwahrnehmung und Raumkogntion Verhalten,
neuronale Mechanismen und Evolution
Hanspeter A. MallotFakultät für Biologie
Institut für Neurobiologie
Kontaktstudientag für Lehrerinnen und Lehrer an
allgemein bildenden Gymnasien und beruflichen
Schulen13.10.2009
2
Übersicht

• Der Kognitionsbegriff
• Typen räumlichen Verhaltens
• Neuronale Grundlagen
• Räumliches Wissen und die Evolution von Kognition

3
Bistable Perception The Schiller Illusion P.
von Schiller, Psychological Research 1930S.
Anstis, Perception, 1980

• Perceptional switching is called cognitive,
  because it cannot be explained as a
  stimulus-response schema.
• The additional cause is called a mental state,
  in this case a percept.
• Other classes of mental states include
  attention, goals, plans, knowledge (beliefs), and
  emotions.

4
Inner State Machine, Level IStereotypes
Input-Output Relation
Flexibility (evolution) requires phylogenetic
time (generations).
5
Inner State Machine, Level IIMultisensory
Integration
Flexibility by alternative stimuli
6
Inner State machine, Level IIIAdaptation and
Associative Learning
Flexibility (learning) requires ontogenetic time
(days to weeks).
7
Inner State Machine, Level IVDeclarative or
Explicit Memory
Goal-dependent flexibility from memory
("insight") achieved in real time (seconds)
8
Summary Cognitive Science as the Science of
Mental States

• Machine theory inner states allow more
  complex behavior
• Observables
• neural correlates
• behavior and performance
• loss of competences in neurological
  disorders
• Cognitive Modelling
• symbol manipulation
• computational theory
• neural networks and control

9
Routes are Stimulus-Response Chains
www.tuebingen.com/rundg/index.html
10
Routes are Stimulus-Response Chains
Go left !
Go right !
11
Routes are Stimulus-Response Chains
12
Routes and Maps
In order to reach the botanical garden, walk
left. In order to reach the university library,
walk right.
13
Routes and Maps OKeefe Nadel, The
Hippocampus as a Cognitive Map, Oxford 1978
Go left !

• Map
• Goal-dependend flexibility
• Latente learning (goal independent)
• Transfer of knowledge to novel routes (shortcuts
  and deviations).

• Route
• Chain of decisions stored as stimulus-response
  schemes.
• Specifically learned for one goal. No transfer
  to novel routes

14
Cognition and Flexibility
15
Übersicht

• Der Kognitionsbegriff
• Typen räumlichen Verhaltens
• Neuronale Grundlagen
• Räumliches Wissen und die Evolution von Kognition

16
Spatial Behavior Memory Involvement
course control no memory required
path integration spatial working memory
visual homing landmark recognition
route memory stimulus-response chains
cognitive maps declarative memory
17
Spezifität und Inaktivierung des Spurpheromons
von Lasius fuliginosus Latr. und Orientierung der
Arbeiterinnen im DuftfeldW. Hangartner, Z.
vergl. Physiol. 57 103-136 (1967)
Worker of Lasius fuliginosus with antennae
inverted

• Trail pheromone sensed by antennae
• Left normal trail following by "tropotaxis".
• Middle left antenna removed, trail following
  degraded
• Right antennae crossed. Trail following by
  temporal comparison.

18
Centering BehaviorSrinivasan MV, et al. J. Expt.
Biol. 199237-244, 1996

• Bee flies in center of corridor with striped wall
• Does not depend on stripe density
• If one side wall is moved forward, bee shifts
  toward this side
• If one side wall moves backward, bee shifts to
  other side.

19
Biomimetic Centering BehaviorArgyros AA,
Tsakiris DP, Groyer C IEEE Robotics and
Automation Magazine, Dec. 2004, 21-30

• Top left robot trajecory
• Top robot at positions 1-4
• Left snapshots from panoramic camera

20
Speed perception fogs up as visibility drops(RJ
Snowden, N. Stimpson, RA Ruddle Nature 392, 450,
1998)

• Adjustment Experiment Constant speed is
  presented in differend weather conditions. Plot
  shows perceived equal speed. In foggy wether,
  speed is underestimated.
• Perceived constant cruising A speed picked
  during clear weather is to be kept constant under
  changing weather.
• People underestimate speed in poor visual
  conditions.

21
Eyetracking in virtual environments
Obstacle avoidance study with patients
(homonymous hemianopia)
G. Hardiess, E Papageoriou, U Schiefer, H Mallot
22
Vermeidung dynamischer Hindernisse die Rolle von
Augenbewegungen
Kreuzungs-Paradigma Rekonstruktion
Aktives Sehen Kollissions-relevante Fahrzeuge
werden besser erinnert
G. Hardiess, S. Storch
23
Spatial Behavior Memory Involvement
course control no memory required
path integration spatial working memory
visual homing landmark recognition
route memory stimulus-response chains
cognitive maps declarative memory
24
Path Integration in Desert Ant Cataglyphis(R
Wehner J. Comparative Physiol. A 189579-588,
2003)
Cataglyphis fortis
Sahara landscape

• Ant lives in salt pans of the Sahara desert.
  Landmarks are absent and pheromones evaporate.
• Ant makes excursions from nest to find food.
• Returns to nest along shortest path (Fig A).
• When transposed during return, runs parallel to
  original path until expected goal is reached
  starts search.
• Fig B one-way-barriers (ramp on nestward side,
  cliff on foodward side) cause deviation during
  return. At the barriers edge, return is resumed
  following the home vector.

25
The Ant Odometer Stepping on Stilts and
StumpsWittlinger M, Wehner R, Wolf H. Science
3121965-67, 2006

• Ants' legs shortened or extended
• Data A Collect ant at feeder, manipulate leg
  length, release, measure run length to nest
• Running distance varies with leg length
• Data B Ants with same manipulation outbound and
  inbound run correct distance
• Ant "counts steps" for distance estimation

26
Path integration in fiddler crabs Uca
vomeris(Video and pictures courtesy of Jochen
Zeil, ANU Canberra)
Female, equal claws
Male, unequal claws
27
Triangle completion with and without
landmarksRiecke B, van Veen H, Bülthoff H (2002)
Presence 11443-473

• Path integration without vestibular input
  (subject seated in front of video projection
  screen)
• Small solid ellipses homing error with landmark
  information
• Large dotted ellipses scene structure for optic
  flow only ("landmark swap", random blobs)

28
Pathintegration recruits the right hippocampus

• Guided movement along two legs of a tri-angle.
  Second leg length and angle varies.
• Behavioural task Point back to start point
• Control in egocentric frame, point to a
  direction indicated at start.

Within subject analysis In right hippocampus,
pointing accuracy correlates with BOLD activity.
Behavioral Performance
Wolbers et al., J. Neurosci. 279408, 2007
29
Spatial Behavior Memory Involvement
course control no memory required
path integration spatial working memory
visual homing landmark recognition
route memory stimulus-response chains
cognitive maps declarative memory
30
Simple place memories
Philanthus triangulum with Apis
melificahttp//www.rutkies.de/grabwespen-opfer
Tinbergen Kruyt, Z. vergl. Physiol. 1938

• Basic paradigm
• nest entrance marked with pine cones
• displace cones after wasp leaves
• returning wasp searches at displaced position

• Identify preferred landmarks
• mark nest with landmarks of two types (A, B)
• displace A and B jointly (control)
• displace A and B to opposite directions
• wasp prefers elevated from hollow objects

31
How honey bees use langmarks to guide their
return to a food sourceBA Cartwright, T Collett,
Nature (1982) 295560-564
Top configuration with three landmarks and two
marked positions Bottom inner circle shows
pano-ramic snapshot obtained at position b, outer
circle shows current image visible at positions a
and b, resp.
Snapshot algorithm. Features (black sectors) and
gaps are matched. If, in current image, feature
or gap is smaller than in snapshot, approach
vector is activated. If position is wrong, turn
vector is activated. Agent moves in direction of
average vector.
32
Snapshot-based homing in humans

• Subject with HMD in 5.2 x 6 m tracked walking
  arena
• Circular room with homogeneous color gradient
• Task
• Subject at position 1
• View scene at position 2
• Walk to position 2
• View scene at position 3
• Dependent measure trajecory, homing error

Gillner, et al., Cognition, 2008
33
Sample Trajectories and Viewing Directions

• a-d subjects head towards goal from the
  beginning
• right heading directions
• top small room, subject quickly finds goal
  directions
• bottom large room, subjects looks around and
  then starts moving towards the right direction.
• General performance is good. Residual homing
  error is well below chance level.

Gillner, et al., Cognition, 2008
34
Homing Performance and Color Modulation
homing in 6 subjects, 4 repetitions
prediction from squared image difference algorithm
Color modulation 10
Color modulation 100
Gillner et al., Cognition 2008
35
Spatial Behavior Memory Involvement
course control no memory required
path integration spatial working memory
visual homing landmark recognition
route memory stimulus-response chains
cognitive maps declarative memory
36
Route vs Map NavigationHartley T, Maguire EA,
Spiers HJ, Burgess N (2003) Neuron 37877-888

• Top Wayfinding condition
• navigate to various goals
• Requires map knowledge
• Bottom Route condition
• Repeat route
• Requires route knowledge
• Control (not shown)
• Follow moving marker
• Requires visual and motor processing

37
Route vs Map NavigationHartley T, Maguire EA,
Spiers HJ, Burgess N (2003) Neuron 37877-888

• Top Wayfinding vs route following
• bilateral perirhinal activation
• in subjects with better performance, difference
  between activities W-R is larger
• Bottom Route following vs wayfinding
• right caudate activation
• in subjects with better performance, difference
  between activities W-R is lower
• Conclusion good navigators recruit caudatum in
  route following and perirhinal cortex (
  hippocampus) in wayfinding condition

38
Skalierbares Modell des Ortsgedächtnisses

• Wegintegration
• Ortserkennung
• Kompassleistungen
• Routen (Assoziiere Orte mit Motorroutinen oder
  Suchaufgaben)
• _______________________
• Topologische Navigation Netzwerke von Orten und
  Übergängen
• Abstände und Winkel
• Routenplanung
• Kommunikation über Raum

walk 2 km
(x2,y2,)
(x1,y1)
39
Übersicht

• Der Kognitionsbegriff
• Typen räumlichen Verhaltens
• Neuronale Grundlagen
• Räumliches Wissen und die Evolution von Kognition

40
Hippocampal Place Fields in the Rat
Rat position at time of spike marked by dots
Rat position tracking in 8-arm maze
C. Hölscher
41
Hippocampal Place Field from Rat CA1J. O'Keefe
et al. Phil. Trans. Roy. Soc Lond B 353, 1998

• rat searching for food on a 40 x 40 cm open
  platform
• 15 CA1 neurons recorded simultaneously
• spike form for each of 4 tetrode channels and
  peak spike rate shown on right

42
"Path integration and the neural basis of the
cognitive map"McNaughton BL, Battaglia FP,
Jensen O, Moser EI, Moser M-B. Nature Reviews
Neuroscience 7663-678 (2006)
Landmarks and path integration Place fields in
the dark
In the vicinity of the box, place field activity
relies on path integration. The range of path
integration is larger if travel is carried out
in the dark. (Gothard et al., J. Neurosci. 1996)
-- box frame
-- lab frame
population vectors
43
"Path integration and the neural basis of the
cognitive map"McNaughton BL, Battaglia FP,
Jensen O, Moser EI, Moser M-B. Nature Reviews
Neuroscience 7663-678 (2006)
1D attractor map model for head direction cells

• Ring topology for head direction cells
• Current heading represented as peak of
  activity on ring
• Upon head turning, the activity peak is
  shifted due to vestibular, proprioceptive,
  and visual input.

44
"Path integration and the neural basis of the
cognitive map"McNaughton BL, Battaglia FP,
Jensen O, Moser EI, Moser M-B. Nature Reviews
Neuroscience 7663-678 (2006)

• Local metric map, location represented as peak
  of activity
• Stack of layers for different headings
• Activity moves according to egomotion input.

Extension to limited 2D field
45
"Path integration and the neural basis of the
cognitive map"McNaughton BL, Battaglia FP,
Jensen O, Moser EI, Moser M-B. Nature Reviews
Neuroscience 7663-678 (2006)
Extension to 2D periodic field

• At margins, local metric map is extended
  periodically.
• The representation of space thus has the
  topology of a torus.
• As the rat moves, the activity peaks keeps
  coming back to the same cell.
• Toroidal local map generates behavior observed
  in grid cells.

46
"Microstructure of a Spatial Map in the
Entorhinal Cortex"T. Hafting, M. Fyhn, S.
Molden, M-B Moser, EI Moser. Nature 436801-806,
2005

• Recording site in entorhinal cortex (red dot)
• results from 3 cells
• left trajectory plus spike locations
• middle firing fields
• right autocorrelation of firing fields
• superposition shows that firing fields are
  interlaced.

c
47
Übersicht

• Der Kognitionsbegriff
• Typen räumlichen Verhaltens
• Neuronale Grundlagen
• Räumliches Wissen und die Evolution von Kognition

48
Phylogeny of Spatial Cognition
man
rodents
birds
fish
molluscs
crusta-ceans
insects
spiders
49
Domains of Cognition

• Is spatial cognition a separate "domain" of
  cognition?
• If yes, how does it relate to other domains
  (objects, causality, social, events and agency,
  language, etc)

50
Core Knowledge of Geometry in HumansS. Dehaene,
V. Izard, P. Pica, E. Spelke (2006). Science
311381-384

• Multiple choice test (odd man out) to assess
  geometrical intuitions in Mundurukú participants
  (Amazonian indigene group without school
  education).
• 6-image slides with one deviating image
• Numbers give correct and mean response time
• Chance level 16.6, 28 35

51
"Core Knowledge of Geometry in Humans"S.
Dehaene, V. Izard, P. Pica, E. Spelke (2006).
Science 311381-384

• Multiple choice test (odd man out) to assess
  geometrical intuitions in Mundurukú participants
  (Amazonian indigene group without school
  education).
• 6-image slides with one deviating image
• Numbers give correct and mean response time
• Chance level 16.6, 28 35

52
Performance of Mundurukú participants in the map
test

• Spatial layout of the experiment. On each trial,
  the participant was shown a "map" representing
  the location of three boxes, one of which
  contained a cue symbolizing a hidden object. Then
  he or she turned around and searched for the
  hidden object (distance d 1.5 m).
• Performance in the three map orientations. The
  map and the environment could be aligned as seen
  from the top, aligned relative to the participant
  before and after turning, or nonaligned.
• Performance at each location in the four
  successive blocks of the experiment.

53
Performance of Mundurukú participants in the map
test

• Performance of US subjects improves with age,
  probably due to school education
• Level of difficulty of individual tasks
  correlates between Mudurukú and US population,
  both children and adults
• Piagetian sequence of topology gt Euclidean
  gemometry is not appearant
• Visuo-spatial and navigational space behave
  equally

54
Can Language Restructure Cognition? The Case for
Space.Majid A, Bowerman M, Kita S, Haun DBM,
Levinson SC, Trends Cogn. Sci. 8(3) 2004

• A number of languages have been identified that
  use an absolute frame of reference ("There is an
  ant on your south leg.") Examples include Guugu
  Yimithirr (Australia), Tzeltal (Mexico), Balinese
  (Indonesia).
• After watching a toy man walk on table 1,
  subjects are asked to indicate the path on table
  two. Speakers of Dutch use relative (egocentric)
  frame of reference while speakers of Tzeltal use
  absolute FoR.
• Conclusion Cognitive categories may be
  language-specific rather than universal

55
Cognitive cladistics and cognitive override in
hominid spatial cognitionHaun DB, Rapold CJ,
Call J et al. PNAS 10317568-17573, 2006

• Human subjects watch experimenter hide a reward
  under a cup at table 1
• Subjects are then moved to table 2 and allowed to
  look under the "equivalent" cup (feedback given)
• 3 conditions (egocentric, object-centered,
  geocentric) in blocks of 10 trials
• Dutch adults and school children (ca 8 years) do
  better in egocentric condition, Haiom adults
  and children do better in geocentric condition.
• This result is consistent with the predominant
  use of frames of reference in the two languages.

56
Cognitive cladistics and cognitive override in
hominid spatial cognitionHaun DB, Rapold CJ,
Call J et al. PNAS 10317568-17573, 2006

• Simplified design with 3 cups and only two
  training conditions (egocentric and allocentric)
• Test with orangutan, gorilla, chimpanzee and
  humans (European pre-schoolers)
• All do better in allocentric condition
• No main effect of species or interactions with
  species
• Allocentric frame of refrence may be common to
  hominidae (plesiomorph). Language-related
  variation in humans is younger addition.

57
Spatial and Other Forms of Cognition

• Munduruku study task difficulty scale is equal
  for Munduruku and US Americans, as well as for
  adults and children gt spatial cognition is "core
  knowledge" independent of culture
• Tzeltal etc studied prefered frames of reference
  in spatial reasoning depend on culture

58
Übersicht

• Der Kognitionsbegriff
• Typen räumlichen Verhaltens
• Neuronale Grundlagen
• Räumliches Wissen und die Evolution von Kognition