Cognitive Science > Action and Cognition > Lecture/8
Contents |
[edit] Questions
[edit] Q3: What limits temporal and spatial resolution?
- Temporal: Delay (blood needs certain time to flow) + jitter (some capillars slower, some faster) ~ 500ms
- Spatial: Resolution limited by ventricles, not neurons (more than one neuron per ventricle)
[edit] Ingredients
- Big coil for magnetic field
- Big coil for electric field
- Smaller coil for making changes to electric field
[edit] Q6: Experiment
- If luminance contrast contributes to fixation points, increased fixation
- Compare with other control points, technical difficulties
[edit] Color
- Electromagnetic waves hit object
- Part of waves are reflected (part of spectrum)
- Reflectance spectrum: Gives for each frequency the fraction of incoming light that is reflected
- Incoming power spectrum .* reflectance spectrum = spectral power distribution
- Different illumination changes spectral power distribution greatly
- Color invariance: Same color assigned to object although different spectrum
- Red/green sensitivity of receptors very close together in humans
- Relative activation of M and L cones is invariant to shadows
- No shadows in M/L difference picture -> find actual objects
- Mathematics
- Response = Integral over all wavelengths: stimulus(wavelength) * response(wavelength)
- Pointwise multiplications (incoming spectrum * reflectance ...) is linear
- Differentiation of colors
- Cones do not see color: Strong stimulus * weak sensitivity == weak stimulus * strong sensitivity
- Different response for different cones
- Colors as relative responses (ratios)
- Color blindness
- One type of cone missing -> projection on lower-dimensional space (2D)
- Makes it impossible to distinguish some spectra
- Metamers
- Differnt physical spectra may map to same activation
- Projetion from infinite-dimensional space to three-dimensional subspace
-
Analyze all spectra with physical device, PCA to identify most common spectra
- E.g. Six-dimensional space can capture most of variance in environment
- Synthesis
- Responses of three cones are not orthogonal
- i.e. they overlap
- Therefore, measuring the activation of a cone type and reproducing stimulus at measurement wavelength is not enough
- This would result in activation of other cones too, not only measured cone
- The same goes for other receptors (camera, scanner)
[edit] V2 afferents and efferents
- Orderly projection from V1 onward
- Neurons in interblob regions oof V1 project to interstripe regions of V2
- Neurons in interstripe region of V2 project to V4
- Neurons in IVB of V1 project to thick stripes of V2, then to V3 and V5
[edit] V4
- Gate to the ventral pathway
- Much smaller than V1 and V2
- Reverse correlation of stimuli types in movie to determine active regions
- V4 afferents and efferents
- Input not only from blob regions of V1 and thin-stripe regions of V2 but also from interblob regions of V1 and interstripe regions of V2
- Therefore: Not only color processing
- V4 response properties
- Achromatopsia
- Bilateral damage to V4 (lingual and fusiform gyri)
- Inability to identify or discriminate gyri
- Colorblind painter: Car accident, no clear damage; not only color processing removed, but also color memories
- Strong attention effects: Increased tonic part of response if attention on object
- Explained by experiment: Visual effects of FEF microstimulation on V4
- Behavioral attention generates effects in V4 that are identical to subthreshold activation in FEF
- Selectivity to 3D
- Acitivity dependent on 3D shape (even if 2d-projection of stimuli looks the same)
- Strong dependency on tilting angle of stimulus
- Makes sense: Gateway to ventral stream (form etc)
[edit] Video: Attention, object recognition, change blindness (person behind counter changes, 75% dont notice)
[edit] FEF
[edit] Frontal eye field
[edit] Eye movements can be elicited with very low current
[edit] Connectivity
- Heavily interconnected in topographic fashion with areas in dorsal and ventral streams of extrastriate cortex
- Topographic: Nearby stimuli -> simiar saccades
- Two differnt systems for saccades of small and large amplitude
[edit] Attention response function
- Experiment: Large number of colored circles, task: Track a subject
- More activity of FEF if attentive than if not attentive
- Differentiate between task-dependent, load-dependent and intermediate regions by parametric task (number of items to be tracked)
- Task regressor > load regressor (difference task/no-task greater than linear dpendence on task difficulty)
[edit] Task-specific vs attention/load-dependent regions
[edit] Task-specific
- Gain in activation between passive and active conditions, no additional gain
- Basic support functions like planning a saccade
[edit] Attention-specific
- Load-related increase
- e.g. Working memory
[edit] Experiment: Visual effects of FEF microstimulation on V4
- FEF microstimulation, causes saccade
- V4 recording that is at end point of elicited saccade in FEF
- Increase in spike rate after onset of visual stimulus (known)
- Subthreshold activation in FEF (500ms after visual stimulus)
- No real saccade
- Activity in V4 is now larger than without microstimulation in FEF
- Effect specific to preferred stimulus of region in V4 -> not just spread of electricity
- Subthreshold stimulutaion facilitates response-property-specific and topography-specific the processing by V4 ===