Cerebellar Function and Dysfunction in Attentional Processes
Methods, Findings, and Prospects with fMRI and EEG
- 2 parts to this talk: autism (EEG & structural MRI), normal cerebellum (fMRI)
- autism as a model system for developmental cerebellar dysfunction
- `clean', uniform lesions
- large, cooperative group of patients and families
- behavioural symptoms
- autistic behaviour as epiphenomenon
- MR cerebellar abnormality
- deficiency of Purkinje cells
- micrograph of cerbellar cortex
- adults: depletion & shrinkage in roof nuclei & olive; shrinkage in dentate
- children: normal populations in all nuclei, but large somata
- no retrograde degeneration of olive; maybe maintenance of foetal olivonuclear projection - would explain differences between children & adults
- behavioural cerebellar signs
- the auditory-visual shift experiment
- lesions: resections of cerebellar astrocytomas
- 4 in lateral hemispheres
- 1 in superior cerebellar peduncle
- auditory-visual shift behaviour
- the Sd wave
- a difference of difference waves:
- (Shift Hits - Shift Ignores) - (Focus Hits - Focus Ignores)
- in terms of basic operations, (detect+shift - ignore) - (detect - ignore)
- similar to Harter's EDAN & LDAP (where central cue directs attention laterally)
- replicated in visual-visual (form-colour) shift experiment
- the Posner experiment: shifts between spatial locations
- ERP index of statically allocated attention: the P1
- The fast shift experiment combines the dynamics of active shifting with the temporal specificity of EEG
- Two `oddball' paradigms running side-by-side; a target in the attended location cues an overt, behavioural response and a shift to the opposite location
- ambiguity in mapping responses to targets: response window for a missed or ignored target overlaps response window for a correctly detected target in the same hemifield
- constrain T+Dtclose <= T+ti+ti+1+Dtopen
- Use Markov chain to implement ti+1 >= Dtclose-Dtopen-ti
- normal subjects: 5 women, 7 men, right-handed, normal or corrected vision, ages 17 to 30
- autistic subjects: 8 men, ages 19 to 32; DSM3R, CARS, ADI, ADOS Dx of autism; 5 P-, 3 P+; no Fra(X); 7 RH, 1 LH
- electrodes arranged in a Hjorth array, 4 central recording sites
- phase-locked and non-phase-locked averages
- signal processing steps
- replicated previous behavioural results
- normal subjects: phase-locked responses
- first 300ms: RH responds to LVF & RVF targets; LH responds primarily to RVF targets
- latency decreases with increasing inter-target interval
- longer latencies
- LH bins 0,1,2 amplitude greater to left targets (rightward shifts) with trend toward decreasing latencies
- RH bin 0 amplitude greater to right targets (leftward shifts), but RH bin 1 amplitude greater to left targets (rightward shifts)
- if given enough time to implement the shift, RH becomes involved in attention to either hemifield
- phase-locked modulations reflect spatially specific enhancement of neural processing
- normal subjects: non-phase-locked responses
- RH: lesser to left targets, then lesser to right targets; LH: lesser to right targets, then lesser to left targets
- complementary, non-phase-locked modulations reflect functionally specific enhancement: increase in processing of target stimuli at the expense of non-task-related processing
- 300..600ms inversion similar to Sd & LDAP
- autistic subjects: phase-locked responses
- response always greater to left targets, in both hemispheres
- autistic subjects: non-phase-locked responses
- response always greater to left targets, in both hemispheres
- failure of spatial and functional specificity
- fixation issues
- in the autistic group, the 2 particularly good fixators matched pattern of entire group
- despite greater magnitude of eye movements in autistic group, there was less modulation of EEG
- interpretation: in autism, normal mechanism of attention fails to operate - slower shifting is a consequence of an ersatz mechanism involving nonspecific orienting
- subjects, tasks, and imaging parameters
- regions of interest
- echo-planar images are warped horizontally
- unwarp by registering each scan line to anatomical image acquired during same session, then iteratively smooth the resulting vertical vector of displacements till it becomes monotonic
- activations: double dissociation of attention and motor functions
- 3D reconstruction
- time series
- What is it that the cerebellum does?
- other studies
- cortex active during tuning, deep nuclei during execution
- questions for future study
