Flashcards in Spatial representation and navigation Deck (24)
Distal (long range) vs proximal (short range) cues
Egocentric (relative to the individual) vs allocentric (relative to the environment) frames of reference
Beacons (Pavlovian approach – automatic reflexive behaviour – becomes attractive due to conditioning) vs landmarks (used as a reference)
spatial learning in lab rat
If there is any way of solving a learning task by treating it as a spatial problem, that’s what a rat will do (“position effects”)
Rats can learn complex mazes, e.g. Hampton Court replica (Small, 1900)
Tolman et al’s (1946) “sunburst” maze seemed to show that rats have a sense of direction and can take a shortcut
…though only if there’s a “beacon” available
Train until able to do it quickly
Route used to take blocked off in test
Learnt where goal was relative to starting position – take shortcut
Confound – bright light above the goal – every time close to bright light = food conditioning = beacon
possible reasons for behaviour in the t-maze
Glands brush against floor – mark where they have been
PAVLOVIAN CONDITIONED APPROACH
All possibilities are right – rats does whatever it takes to get the food
odour as a reason
Test in extinction
“+” = trained side but no food there
Swap the arms of the T-maze
If it’s following an odour trail it will go to the “wrong” side.
Still goes the the left side – not following odour trail
Enclosed = learns to go left
Not enclosed – goes to correct side – can see landmarks
landmarks as a reason
If we’ve established that the rat is heading for a certain point in space, the question then arises of how it is doing that. Is it using a map - or just approaching a specific landmark (beacon)?
A - closest – associated – beacon
By taking the landmark away (deletion) we can assess the effect on the rats performance. We can also try altering the configuration (spatial arrangement) of the landmarks
If using configuration – should still go to right location
the radial arm maze
Radial Arm Maze experiments with rats.
There is now abundant evidence that rats typically solve this maze by using the external (extra-maze) landmarks.
Rotation tests and landmark deletion / re-arrangement studies all point to this conclusion.
The animals are forced to the 4 arms shown in blue, then the maze is rotated through 45º (or the landmarks are rotated relative to the maze.
Control = before rotation test them on which arms they go down
Now the rats are offered a choice between an unvisited arm U (animal has not been down it, there is food at the end, but it’s now at the location that was visited) or a visited arm V (animal has just been down it, no food at end, but is at an unvisited location.)
The result is that rats tend to choose the visited arm. In my studies a typical result was that on 32 trials 20 were revisits to previously visited arms. A control test where the maze was not rotated gave only 2 visits to previously visited arms, a highly reliable difference.
Not using intra-maze cues
Going to unvisited place
Suzuki, Augerinos and Black (1980)
used a cylindrical testing chamber with discrete landmarks at each arm of the radial maze.
They found that rats ‘followed” rotation of the landmarks with respect to the maze, but a re-arrangement of the landmarks relative to one another (i.e. transposition) dramatically worsened performance between study and test.
These data suggest that rats use configurations of landmarks to define locations in the radial maze, rather than using them as beacons to mark specific locations close by.
O'Keefe and Nadel (1978)
claimed that animals use a map to navigate, and that the mechanisms for constructing and using this map were located in the hippocampus.
the hippocampus as a cognitive map
There is good evidence that the hippocampus is involved in spatial learning, e.g.
O’Keefe and Nadel: single cell recording in the hippocampus shows “place cells” which fire when a rat is in a particular place in a maze
Hippocampal lesions disrupt performance in the Morris water maze
O'Keefe and Conway (1978)
Some cues may not have been adequately controlled
the Morris water maze
The rat is put in the pool at a random location along the side (S).
It then swims to the platform, at first in a roundabout fashion, later more directly.
Animals with hippocampal lesions are impaired at this task, in that they take longer to find the platform and do not exhibit the ability to swim straight to it in the way that controls can.
works with reversible lesions too
technique can also be applied to humans
blocking in the water maze
a problem for the cog map hyp
Two groups are trained in the water maze to find a platform on the basis of different sets of landmarks, either ABC or ABCX.
Landmark X is then added to ABC in the first group, and more training given.
Tests with ABC and ACX reveal how well the animal has learned to use X to find the platform in conjunction with the other landmarks.
1 = best score
The Blocking Group are worse at using landmark X (see ABX)
Learning is akin to the learning seen in conditioning experiments
navigation: what do you need to find your way around?
A map to specify the spatial relations of objects; cf. Tolman (1948), cognitive map to account for rats’ learning of mazes?
A compass to specify directions (orient the map; cf. Kramer (1953), map and compass account of bird navigation
A locator to tell you where you currently are
Redundant systems so that if one is blocked you can still navigate
magnetic navigation : the Green Sea Turtle (Chelonia Mydas)
Migrate 2000 km from feeding grounds on the Brazilian coast to nest sites on Ascension island
Thought to use orientation and intensity of earth’s magnetic field – a bearing map
Contributions of instinct and individual learning unclear
Sun compass (requires knowledge of time of day). Clock-shift experiments.
- Shift dark light cycle
- Think midnight is midday – think it’s dawn – think it’s east when you know it’s west – fly in wrong direction
- Magnetic field change and grain – no evidence of learning – receive plane/plain? Of polarisation from sun
Infrasound (?) and other beacons
Olfaction (e.g. Guildford et al 1998)
Route marks e.g. motorways (Lipp et al 2004)
Proximal landmarks at start and end of flight (e.g. Biro et al, 2003) – use the way the scene looks when get closer to home – frosted contact lens – cant distinguish parts of scene and can’t find right location
Must all be learned
adaptive influence on spatial learning: scatter-hoarders
Scatter-hoarders make numerous (several thousand) caches of food and recover them months later
Often the look of the environment is different at cache time and recovery time (e.g. snow)
Caches cannot be marked (e.g. by scent) or they would be pilfered, so scatter hoarders need exceptional spatial memory
Examples: corvids (scrub jays, Clark’s nutcracker), sciurids (grey squirrel, fox squirrel), parids (marsh tit, coal tit)
Note not all members of same genera/families scatter hoard
exps on scatter hoarders
Krebs et al.
Kamil et al.
Clayton et al.
trained grey squirrels to find nuts she’d buried at random places in a 2m circle
visual signal when nuts would be present
squirrels could still recover nuts accurately 2 months later
decoy nuts buried at different distances from targets… if decoy was more than 2cm away, the squirrels always took the target
Krebs et al.
compare storing parids with non-storing species on spatial memory tasks
scatter hoarders do better (e.g. Krebs et al 1990)
Kamil et al.
series of experiments on Clark’s Nutcrackers, investigating what cues they use to find caches
e.g. Kamil & Jones (2000) conclude that birds can use both absolute and relative cues, and both distance and direction from landmarks, but direction is more salient
Clayton et al.
many experiments on scrub jays showing that they remember what they have stored and when, as well as where (see previous lecture)