Spatial cognition

Spatial cognition

1. Spatial cognition and navigation: Remembering layouts of objects and where things are. Small scale and large scale. Navigating is crucial - from where is my laptop to getting round a city.

2. Topographical Matching Task/mental rotation test

   • Mental rotation - are shapes the same if you rotate? Truing to do it in head - spatial cognitive task. It is taxing. Topographical - which landscape is the same.

3. child friendly spatial memory task - Duck hiding under a cup. Then move board round. Then a different view. Use landmarks!

4. The emergence of spatial abilities. How do children develop abilities to solve spatial problems? Use coding tasks and manual manipulation tasks.

   • How do children develop abilities to solve spatial problems? Some explanations to consider: Changes in mental representations of space - Frames of reference, Sensory cues used for spatial coding, Basis in the brain. Changes in attention, strategy, understanding

5. Lecture outline: Historical introduction, Conceptual framework: representations, FORs, the brain, animal models, Studies of objects and mental rotation, Studies of layouts and spatial recall, Conclusions

6. 1. Historical introduction

   • Piaget - first to investigate spatial cognitive development.
     • “Three mountains task”. Not solved until 9-10 years (see also video). Confusion of own view with how it would look from another perspective: “egocentrism”. Here is a view of a landscape. Here is a map of the same landscape seen from above. Would you see the view above if you were standing at position A, B, C or D on the map?
     • “Model landscape task”: Show where an object placed on one model would be on the other. 3-4 years – use environmental features but not places on the model. 4-5 years – use a single landmark. 7 years – use full range of landmarks and spatial relations.
     • Piaget - Constructivist theory of cognitive development. Children build up increasingly sophisticated spatial concepts from elementary sensory-motor behaviours. Learning via experience and the senses, in the Empiricist tradition (recall nativist / empiricist lecture). Contrast (briefly) with: Immanuel Kant - 18th C. rationalist philosopher. Influential thesis that some human concepts are universal and independent of experience – including spatial concepts. Ifyou are brain in vat, you would understand space. Dont need experience just need to be rational. Opposite! Elizabeth Spelke – current developmental researcher. Proposal that infants come equipped with “core knowledge” – including some knowledge of spatial layouts and geometry. Core knowledge. Even if born with some, clearly not everything is there, some needs to be learnt. --> See L2 spatial/number lecture on “innate geometry”
     • Subsequent studies of development (preview): It turns out that if we simplify things, we can show some kinds of spatial competence much earlier – but shifts from “immature” to “mature” spatial behaviour still remain e.g. “egocentrism” in the first year.
       • Acredolo, 1978; Acredolo, 1980. After 1 year, switch to look to opposite side.

7. 2. Conceptual framework: representations, FORs, the brain, animal models

   • How can we explain developmental changes in spatial coding? Representations: what kind of spatial information the brain is encoding and storing.. In Piaget’s “model landscape”: features --> one landmark --> multiple landmarks and distances. In Acredolo’s orienting task: -->direction relative to self --> direction in the room. What kind of spatial info is stored and encoded.
     • Representations: what kind of spatial information the brain is encoding and storing. In the case of spatial coding in particular, we need to consider the use of different frames of reference. Remembering where something is means remembering where it is relative to something else…
   • Spatial frames of reference (FORs). Where is the cup? Relative to teapot, bodty etc. Depends on what you want to do.If you want to pick up the cup, the useful FOR is your body, If you want to pur tea, the useful For is the teapot.
     • Interpretation in terms of representations and frames of reference (FORs). Representation of frames of reference. In case with big star to landmark window - you use allocentric. If ecogentric, then get it wrong, and using direction relative to own body
     • Another issue about representations and FORs: It is also possible to maintain an egocentric representation of a location, but update it while moving – i.e. continuously maintain/correct an egocentric representation by accurately tracking own movement (through vision, vestibular system, step counting, etc) --> An alternative account of the representations underlying correct responses in Acredolo task. Also, outward path while foraging. Many organisms, including ants, navigate using very accurate “spatial updating” (AKA “path integration” or “dead reckoning”). Mammals, including humans, can also do this, although they are not precise over large distances. Alternative to using landmarks to solve is to code own movement.
   • Spatial representations in the brain: adult neuroimaging and animal models. The brain has multiple, complementary representations of space.
     • Parietal lobe: Coding locations relative to self (e.g. to reach for a cup – EGOCENTRIC). Disorders: neglect, e.g. after stroke. Reasonably simple but important. Stroke that afects parietal lobe - cant reach or neglect part of space.
     • Medial temporal lobe, including hippocampus. Flexible navigation in space (ALLOCENTRIC). Disorders: disorientation (and memory problems), e.g. in Alzheimer’s. Two different systems for coding in space.
     • ‘Place’ vs. ‘response’ learning in rats - Compare infant orienting responses in Acredolo studies
       • Initial learning of ‘place’, subsequent learning of body turn. ‘Place’ learning is dependent on hippocampus (e.g. Morris et al., 1982; O’Keefe & Nadel, 1978) ‘Response’ learning is dependent on basal ganglia (e.g. Potegal, 1972; Kesner, 1993). Basal ganglia: caudate, putamen, nucleus accumbens, globus pallidus. Packard and McGaugh 1996. Plus maze. LEarns which way to get reward. Then put in other side - response learning or place learning? When first - place learning. Then body. Ledion them, depends on dif areas.
       • A neural circuit for self-location/orientation in the rodent hippocampus - place cells. Same in humans
     • Allocentric spatial coding and flexible navigation also dependent on hippocampus in humans
       • Right hippocampus activation correlates with accuracy at a virtual navigation in adult fMRI (Hartley et al, 2003). A patient with hippocampal damage can recall locations from a familiar viewpoint but is impaired given a new viewpoint (King et al, 2002). London taxi drivers have larger hippocampi than controls, and the volume is correlated with time spent as a taxi driver (Maguire et al, 2000)
   • Spatial representations in the brain: adult neuroimaging and animal models. The brain has multiple, complementary representations of space. --> Dissociations, e.g. between brain areas encoding “egocentric” vs “allocentric” information. --> These can also guide our thinking about what might be changing in development. (How?) Hypothesis - allocentric coding takes longer asbrain areas supporting take longer to mature

8. 3. Studies of objects and mental rotation

   • Relevance: object / tool use and manipulation. Have to use objects and tools. Hallmark of humans - ability to use hands, make tools, understand what objects fit
     • Piaget – no mental rotation abilities until 7-8 years. 5 and 8 year olds. Stimuli - bear with dif hands raised. Pretraining, child given same or different bears. Chld had to identify and justify why they were the same or different. Criterion, push lever. If correct the put through. Experimental - bear on left remained upright, tight was rotated. Had to say same or dif. Testing if child used side or which arm. Found both 5 and 8 year olds used same ways to mentally rotate. Measured the reaction time - presume getting it mainly right. 5 yr olds- longer reaction time for longer rotation. 8 year olds quicker, but still a correlation. more errors for 5 years old. Marmor (1975, 1977) – earlier onset, 4-5 years
   • Looking time experiments (as is often the case) – even earlier evidence for “mental rotation” abilities – in the first year. Hespos & Rochat (1997). first to show. Möhring & Frick (2013) --> “Success” (sensitivity to possible vs impossible) at 10 months. Chidren understand spatial rotations and the biomechaniscs of hand. Predicted by individual motor development (via questionnaire) – importance of action experience for cognitive development. In line with Piagetian ideas – although much earlier competence via looking than he found in explicit tasks. SupportsPiaget that it is through experience you are good. But are those children specifically advanced on motor development or just better developed in general?
   • From looking to action – beginnings of tool use. Örnkloo & von Hofsten (2007) – and thesis. Solved the problem of correctly orienting objects to fit into holes at 22 months (making appropriate pre-adjustments). Quite a while after reaching and handling objects
   • Individual differences, including in later development – see Newcombe et al (2013). Correlation with motor experience (e.g. as in Möhring & Frick, 2013). Gender differences (e.g. male advantage from 4½ years, Levine et al, 1999). A predictor of success at STEM subjects (e.g. engineering) - see Newcombe et al (2013). Makes sense, probably involves thinking about spatial layouts. NB – traditional differences in approach. object / mental rotation studies – focus on individual differences (more research on underlying representations / mechanisms would be useful). Layout / spatial recall studies (next) – focus on representations / mechanisms (more research on individual differences would be useful) - (Newcombe et al (2013)

9. Studies of layouts and spatial recall

   • We saw that Piaget’s tasks, which had quite complex cognitive demands, were not solved until late in childhood.More recent research has been directly inspired by “animal” tasks with simple responses – e.g. Mangan et al (1994) – find a toy in one of 8 identical hiding places in a circular environment - success at 24 months. Overman et al (1996) – human adaptations of the “water maze” used with rats (ball of styrophome to find chest)- improvements to 6 years (ceiling effect) --> or 10 years (outdoor circle with natural surrounding landmarks) (original water maze described in Morris, 1981)
     • Cue control in animal tasks (and human tasks inspired by them). By manipulating which information is available, we can check which information subjects are using (Conceptual framework: representations, FORs, the brain, animal models – see which kinds of spatial information subjects are representing). Quite good cue control - do we let them use landmakrs, spatial updating etc. By dissociating them, we can see what spatial information they are representing.
       • Use of animal models and tasks – works well in spatial cognition as many tasks are directly comparable (e.g. search). Evidence for comparable spatial coding across many animal species and humans – allows for study of neural bases (e.g. rat model for spatial memory – see earlier section on Brain). Questions about what is shared, and what is unique to humans – for one view, see Wang & Spelke (2002) - Human spatial representation: Insights from animals. A few basic processes are shared between humans and animals, but some like map reading are just humans.
   • Examples of cue control in the water maze
     • Add / remove / rotate landmarks – test use of landmarks. Remove rat, disorient (by turning), return to maze – test use of self-motion. Keeping trakc of own movement - disrupt. Have to rely on landmarks.
   • Use of disorientation to test for use of VISUAL LANDMARKS ONLY: rats
     • After seeing food hidden rats are disoriented. Which cues do they use to re-establish orientation and find the food? Cheng (1986). Rats olds use room shape (geometry) to reorient. But don’t use features and odours to disambiguate corners. Room shape (geometry) is the main “landmark” represented by rats. Rat in rectangular enclosure with distinctive pattern smells, wall colour. Disoriented. Have to use cues. Seem to use geometry but dont use features or colours.Plausible - if you walk round natural environment it stays the same even if trees etc change! Geometry seems to be a good cue!
     • Use of disorientation to test for use of VISUAL LANDMARKS ONLY: humans
       • An equivalent task for 18-24 month-olds searching for a hidden toy. Hermer & Spelke (1994). 18-24 month olds use room shape (geometry) to reorient . But don’t use colour to disambiguate corners. --> Spelke: rats’ basic reorientation mechanism shared with young humans - possibly based on “innate” understanding of geometry? Again, use geometry.
         • Spelke: rats’ basic reorientation mechanism shared with young humans - possibly based on “innate” understanding of geometry? Are some kinds of spatial understanding “innate”? Part of Spelke’s “core knowledge” proposal (Spelke, 2007 – and see nativist / empiricist lecture). Subsequent studies and criticisms: Twyman & Newcombe (2010), Five Reasons to Doubt the Existence of a Geometric Module. In any case, abilities to represent and use additional kinds of landmark (e.g. use of colour landmarks to disambiguate corners) clearly need to develop. NB - cue control: by using disorientation, these studies probe competence at using VISUAL LANDMARKS alone, without the influence of coding relative to the body or path integration
   • Development of spatial encoding when both visual landmarks and body-referenced coding are available
     • --> Sandbox task, Huttenlocher et al (1994) Newcombe et al (1998)
       • Experimenter hides toy in sandbox. Mum has eyes covered so does not see. Child sees toy being hidden. Cond 1: free area with landmakrs. Cond 2 : curtain round sandbox so no external landmarks. Child turns away and then back. Child and mum either moved to other side. Asked where toy. Older generally could, over 21 months, landmarks made a difference! Either retrieve from same side (Representations / FORs: Can solve simply by encoding relative to body) or opposite side (Must update body encoding to take own movement into account OR use external landmarks).
         • Results: 16-36 mo olds are above chance from both “same” and “opposite” side - Can do both simple coding rel to body as well as updating of this or use of landmarks. Adding extra landmarks in room improves “opposite” performance after 22mo, - Some use of landmarks to solve the task from 22 mo. Follow-up questions: How do the body- and landmark-based systems interact when both are available? At what age can children find objects using only visual landmarks? --> disorientation studies showed partial success at 18-24 mo – use of geometry in particular but not the full range of landmarks. --> Piaget’s model landscape suggests ~7 years – but not a direct search task
     • Nardini et al 2006 - to try and dissociate landmarks and self spatial updating. Duck under cup paradigm. See toy hidden, walk away, then come back. Either same viewpoint - toy keeps its place relative to the body, the room and the board, so can use simple egocentric coding, updating, landmarks in room or landmarks on board. They progressively took cues away. Then, different viewpoint, walk 135degrees around board.. The toy has not kept its place relative to the body. But has to the room and the board, so can stil use updating, landmakrs in room and landmarks on board. Then rotated board by 135d. Toy only keeps place relative to board, so need to use landmakrs on the board. Have walked back to same place so your movements are no guide.
       • Results - Spatial frames of reference develop at different rates: room f.o.r. by 3 years (spatial updating and/or use of landmarks in the room) . Evidence for board (local landmarks) only at 5 years (allocentric coding using the local landmarks within the array). Late – but earlier than in Piaget’s task – direct search, more easily understandable.
   • More questions: i. How should we interpret failures to use the local cues below 5 years? Changes in mental representations of space – not representing relative to board landmarks at <5 years - basis in the brain? e.g. use of landmarks from new viewpoints is associated with the hippocampus – hippocampal development underlying this ability? Or Changes in attention, strategy, understanding. Younger children may also represent relative to board landmarks but use incorrect, egocentric representation to respond. Follow-ups: neuroimaging, individual differences – correlations with other task. Potentially could use the right response but dont choose to use it, hard to test.
     • ii. The more FORs children had available, the better they did (from 3 years) – does this mean that they are combined to improve performance? --> no, more likely the disadvantage when given fewer cues is because Ss select the wrong cue. Nardini et al (2008) - Navigation task in a darkened room. Testing use of landmarks alone, spatial updating in the dark alone, both together. compare with mathematical model predicting advantages in accuracy from combining the two. Adults benefit from combining information as the model predicts, 4-8 year olds do not. --> Not yet combining sources of information to be more precise (see also multisensory lecture).
   • Summary: Tasks with good cue control, often adapted from animal literature, allow us to dissociate representations, frames of reference, and sensory cues used to solve spatial problems. Major changes in these – e.g. coding relative to landmarks later than relative to self. Coding by combining cues later than by individual cues. Interpretation: flexible, allocentric spatial representations take longer to acquire than others. Brain basis – hippocampus? Or: development of attention, strategy, cue selection? More individual differences research would help to resolve.

10. Conclusions

    • Spatial cognition is a crucial part of adaptive behaviour and human intelligence. Piagetian approach – spatial cognitive abilities depend on sensorimotor experience. Analysis of representations and frames of reference; neural representations and animal models. Object / rotation studies – significant development, individual differences including those associated with motor experience. Navigation / recall studies – also significant development, changes in use of cues and FORs.