Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
Top-down & bottom-up control: attention
Transcript of Top-down & bottom-up control: attention
This necessitates control mechanisms that select
information in accord with current behavioral goals!!
(2) We see by learning to see, by finding relationships
between inputs and behavioral meaning
Therefore, behavioral context is everything!! What we see.. ….is very much driven by behavioral context ….is very much driven by experience 3. How is attention implemented? Hopfinger et al., 2000 2. Passive brain? Reflexive, automatic behavior = feedforward processing
Intentional, voluntrary behavior = feedback processing Big changes produce N2PC in lateral occipital complex (LOC), which has large receptive fields.
Small changes produce N2PC in V4, which has relatively smaller receptive fields.
The neural locus of spatial attention thus seems to move through the system depending on the required size of the receptive field. A. Effects of attention (in visual system) 1. Increases in neural activity at a specific spatial location 2. Increases in neural activity are weaker lower in hierarchy Hopfinger et al., 2000 Tootell et al., 1998 Tootell et al., 1998 Somers et al., 1999; PNAS Latency of attentional modulation of firing rate in areas V1, V2, and V4. (A–C) Red traces represent spike density plots of the average response in each area with attention directed INTO the neuron’s RF (as illustrated by the “spotlight” of attention in the drawing in the top panel of F). Blue traces represent responses with attention directed OUT of the RF (F, Bottom). Responses are shown for area V1 (A), V2 (B), and V4 (C). Responses were aligned to the stimulus onset (0) and were smoothed with a Gaussian window of 30 ms. Shaded areas represent SEM. Vertical black lines represent the onset of the attentional modulation in each area. (D) Distributions of the latency of the attentional modulation are shown for each area. Red: V4; Blue: V2; Green: V1. Arrows denote median latencies for each visual area. (E) Cumulative distribution plot of the latency of attentional modulation in each area; colors represent the three areas as in the previous plot. (F) Cartoon depicting the stimuli in the blocked design task. 3. Neural site of attention is flexible (spatial attention) Hopf et al., 2006; JON 4. Neural site of attention is flexible (feature/object) 6. Attention improves tuning of neurons Main task of cognition is to guide action
Success of cognition is therefore NOT the ‘correct’ representation of environmental features, but the generation of actions that are optimally adapted to particular situations.
Internal states: not primarily stimulus representations, but stimulus representations modulated by topdown factors (attention, working memory, behavioral context). These help decide what to do (“action pointers”) Each percept is synthesized from elementary bits of information.
Representational contents of neurons increase incomplexity as one moves up the processing hierarchy
Brain is a passive, stimulus-driven device
Serial, “bottom-up” processing in hierarchically organized neural architectures: a. Classical theories (the feedforward model) (1) Similar timing of activation different stages within the
• V1, MT, FEF: activated almost simultaneously
(2) Feedforward connections are not the sole determinant of
response strength and selectivity of visual cortical neurons
• Importance of the local connectivity network
• Feedback connections b. Evaluating the feedforward model c. Paradigm shift Brain is an active and adaptive system
Information processing always occurs in a context
-> strong influence of top-down factors
(attention, working memory, mood, behavioral context)
Intimate relationship between cognition and action d. So.. what about cognition then?? ‘Manteldiertjes’: Eat their brains after
they, as grown-ups, attach to a rock
and will no longer have to move around Do we need a brain? Or active brain? 5. Attention resolves competition Moran & Desimone, 1985 Desimone et al., 1995 Kastner et al., 1998 Martinez-Trujillo and Treue (2004), Current Biology Buffalo et al., 2010 CRITICAL NOTE
Sometimes, effects of attention are not found in early visual areas (especially in electrophysiology studies). This may have multiple reasons.
1) Low temporal resolution of fMRI: V1 modulation may occur later, and be driven by extrastriate feedback
2) Differences in experimental paradigms: Electrophysiological studies showing clear effects used difficult tasks with low-contrast stimuli
3) Attention is efficient: it only recruits lower areas in hierarchy if necessary
4) Differences between subjects Instruction: Attend to motion or colour
Instruction evokes activity in MT (motion in red) or in V4 (colour in blue). Schoenfeld et al., 2007; Cereb Cortex O'Craven et al., 1999; Nature Instruction: Attend moving object or static object.
Instruction evokes activity in FFA or in PPA 3.The mechanism behind top-down attention: Synchrony B. The brain circuitry of top-down & bottom-up attention 1. The suspects The dorsal network = top-down control of attention
The ventral network = bottom-up attention ("break") Pulvinar = engaging attention Superior collicoli = move attention Frontal eye fields (FEF) & Intraparietal sulcus (IPS):
1. send down cortico-cortical feedback to visual cortex
2. recruit the superior collicoli (move) and pulvinar (engage)
At the same time, there is a circuitbreaker:
the temporoparietal junction (TPJ) &
the inferotemporal gyrus (IFG)
break top-down attention when a stimulus captures attention 2. Evidence that the FEF and IPS are involved with top-down attention TMS at IPS and FEF:
1. slows down performance
2. reduces accuracy Capotosto et al., 2009 Temporal binding model:
* Synchrony enhances saliency of neural responses (because correlated discharges have a much stronger impact on neural populations than temporally disorganized inputs)
* Leads to selection and grouping of subsets of neural responses for further joint processing
* Synchrony can be intrinsically generated, and modulated by intrinsic signals that reflect experience, contextual influences, and action goals Engel & Singer, 2001; Singer & Gray, 1995; Varela et al., 2001 Core prediction:
- Relationship ongoing activity and informationprocessing
- Pre-stimulus oscillatory activity patterns should predict
stimulus processing C. Top-down versus bottom-up competition How is perceiving (P+) different from not perceiving (P-)? Calculating inter-site synchrony Pre-stimulus oscillatory activity patterns predict visual detection performance!
Hansmayr et al., 2007 Attention modulates early sensory processing by biasing sensory brain areas in advance to favor certain stimuli over others
This top-down bias is established through feedback connections from dorsal frontal and parietal cortex
Both FEF and IPs can bias sensory processing
Intrinsic neural synchrony between interconnected lower and higher brain regions provides an important mechanisms through which input selection is controlled MECHANISMS OF TOP-DOWN ATTENTION Recommended reading:
The brain circuitry of attention
Shipp et al., 2004
1. Two networks 2. Neglect is caused by a imbalance in the system Dorsal system:
top-down control &
sensory saliency detection
(strongly right lateralized) Competition between dorsal and ventral system.
Dorsal activation decreases ventral activation Required reading:
The Reorienting System of the Human Brain: From Environment to Theory of Mind
Corbetta et al., 2008 Spatial neglect:
Failure to attend to,look at, and respond to stimuli located on the side of space or body opposite to the side of the brain lesion.
Caused by lesions in the ventral, bottom-up attention pathway. Initially neglect is very severe in the acute phase, but eventually it becomes better (chronic phase). It seems that damage to the right hemisphere causes a disbalance in the left hemisphere that normalizes over time. Corbetta et al., 2005; Nature BOX 1: EEG indexes of attention N2PC ERP waveforms recorded from monkey P (A) and monkey S (B) for right (blue) and left (red) visual field targets across all pairs of lateralized electrodes for search arrays with two items. The amplitude of the contralateral positivity was maximal and significant at the most posterior pair of electrodes (P < 0.01) and decreased progressively at more anterior electrodes. For monkey S, both the early positivity contralateral to right visual field targets and the later positivity elicited by left visual field targets show this pattern. This mirrors the scalp distribution of the human N2pc. Set size 2 arrays are shown because they elicited the largest amplitude effects. P1 Paradigm for using ERPs to study attention. Stimulus display (left) and idealized results (right). Subjects fixate a central cross and attend either to the left or right visual field. Stimuli are then presented to the left and right visual fields in a rapid sequence. In this example, the ERP elicited by a left visual field stimulus contains larger P1 and N1 components when the stimulus is attended (‘Attend left’) than when it is ignored (‘Attend right’). Woodman et al., 2007; PNAS (shifts of spatial attention) (locus of spatial attention)