Learning in visual search Flashcards
Contextual cuing
~ when the configuration of the distractors, cues you to where the target of your search is
~ learning guide the attentional spotlight
Recognition performance
~ these are ‘subjective’ measures of your awareness
~ likely to be insensitive to knowledge
~ participants have low confident in their knowledge
~ better method = forced-recognition
Chun and Jiang (1998)
~ first demonstration of contextual cuing - showed quicker reaction times when participants presented with repeated sequences
~ given old-vs new recognition task –> 52% correct but no sig. above chance
~ suggested that contextual cuing driven by implicit learning
Implicit learning
Learning in the absence of any conscious awareness of what has been learnt
Smyth and Shanks (suggestions)
(2008)
~ perhaps recognition tests used previously were not sensitive, two changes:
- generate target quadrant
- repeat testing = more data
Smyth and Shanks (study)
(2008) ~ repeated configurations had a higher percentage correct ~ performance was above chance ~ took time to show this ~ similar retest effect observed ~ supports implicit learning
Jiang and Chun (2001)
~ when attention was focussed on one set of distractors (green) the repetition of the other set (red) went unnoticed
~ unattended distractors did not cue target location
~ selective attention modulates contextual cuing
Computational model of contextual cuing
Output = model needs to decide where to look for a target Weights/learning = model needs to learn how to use the input to produce the correct output Input = model will be given the features of the scene (stimuli)
Brady and Chun’s model of contextual cuing
(2007)
~ the model is learning to predict the correct location of the target form the spatial arrangement of objects in the pattern
~ model has a pattern of activation on it output units
Brady and Chun’s model of contextual cuing - STIMULATION RESULTS
~ the model can produce a contextual cuing effect
~ it checks fewer location on predictive patterns –> learnt to predict where the target is for the patterns
Benefit of modelling
~ can be used to account for contextual cuing
~ can be used to generate clear predictions
~ predictions can be tested and models can be falsified
Computational model
a precise definition of a part of the cognitive process
A theory that is merely expressed verbally…
… can often be presented with a multitude of caveats
–> bad for science