Lecture 11 Flashcards
what are the two areas of visual difference reported across multiple conditions?
- global motion perception (including discussion of Manning et al., 2015)
- Visual stress
what are some ways to have more inclusive and destigmatizing language when discussing sensitive topics?
· Disorder vs. condition
· Person-first (e.g., person with autism) vs. identity-first (e.g., autistic person) – e.g., Keating et al., 2022, Autism Res.
· Respect neurodiversity
· Avoid value judgments and deficit-based language where possible (e.g., “higher thresholds” vs. “worse performance”)
· “Typically developing” preferred over “normal” or “healthy”
what is the Titchener circles task?
presented with two sets of several circles
one set has a centre circle surrounded by seemingly smaller circles
the other set also has a centre circle but is surrounded by seemingly larger circles
the “illusion” is that the surrounding circles make the centre circle look like different sizes (they are not)
how do people with autism perform on the Titchener task?
(center circles are the same size, however sizes of surrounding circles make them seem different)
people with autism are more likely to say they center circles are the same size
begs the question of why don’t they get tripped up on it?
what is the motion coherence task?
Colours and arrows in diagram are just illustrative to show coherently moving signal dots (purple) and randomly moving noise dots (blue)
Make the task harder by reducing the proportion of signal dots.
Signal dots - dots moving in the same direction
Noise dots - dots moving in all directions
100% coherence = 100% signal dots
Trying to figure it out at what percentage of coherence do people start agreeing on a general direction of motion / aka what’s the threshold
when testing people across multiple conditions, what threshold of motion coherence was found?
Higher motion coherence thresholds in a range of conditions, e.g.,
· Williams Syndrome (Atkinson et al., 1997)
· Autism (Spencer et al., 2000)
· Dyslexia (Hansen et al., 2001)
· Hemiplegia (Gunn et al., 2002)
· Fragile X Syndrome (Kogan et al., 2004)
· Schizophrenia (Chen et al., 2005)
· Congenital cataract (Ellemberg et al., 2002 vs. Lewis et al., 2002)
To narrow down on a couple of these conditions: autism and dyslexia, as its quite interesting how differences in motion coherence thresholds have been interpreted quite differently in these two conditions.
how does Carroll et al., 2025 define dyslexia?
difficulty in reading and spelling, associated with multiple factors
what did they Stein et al., attribute elevated motion coherence thresholds in dyslexia to?
atypical magnocellular function
anatomical evidence: reduced size of cells in magnocellular layers in LGN in dyslexia (Livingstone et al., 1991)
only ~30% of dyslexic people show difficulties with motion perception (Gibson et al., 2006)
(suggests it isn’t a cause of dyslexia)
not always adequate (non-magnocellular) comparison tasks
what is a magnocellular system?
carries info with high temporal frequencies - important for perceiving motion
how does the DSM-5, 2013 define autism?
- Differences in social communication and interaction
+
- restrictive and repetitive behaviours and interests, including sensory processing differences
what did Happe et al., attribute elevated motion coherence thresholds to?
difficulty integrating parts into the whole
autistic individuals may pool over a smaller number of dots
what could autistic individuals difficulty integrating parts into the whole be linked to neurally?
Neurally, could be linked to reduced functional connectivity, stronger local than long-range neural connections, reduced top-down modulation (see Happé, 2021).
what is the domain-general theory of autism?
Domain-general theory of autism, which suggests that autism is characterised by a difficulty to integrate parts into a whole – where the world seems ‘fragmented’ and it is hard to pick up the gist and use contextual information.
This is sometimes referred to as “failing to see the wood for the trees”.
It has been proposed as a general cognitive style covering multiple domains, including perceptual, visuospatial, and verbal abilities.
can the weak central coherence account explain reduced susceptibility to visual illusions in autism?
Yes! Its been proposed that what autistic people are doing when they’re looking at these illusions is that they’re able to process the circles individually and they’re not influenced by any context because they’re focusing on those local details
what reasonings did Manning et al., find against research in why motion coherence thresholds elevated in autistic children?
reduced sampling? reduced global pooling over dots (weak central coherence)
higher internal noise? imprecise estimation of each local direction (Simmons et al., 2009)
motion coherence task cannot resolve this
what are the problems in motion coherence tasks according to Dakin & Frith (2005)?
The problem with the motion coherence task is that there are actually multiple reasons why autistic children could have elevated thresholds
We’ve just talked about the Weak Central Coherence account – this would suggest that autistic children have reduced sampling – that they are combining or pooling over fewer dots than typically developing individuals, in line with the weak central coherence account.
But it’s also possible that they are less precise at working out the direction of each individual dot, and if these individual direction estimates are pooled, this will also lead to elevated thresholds – increased neural noise been proposed in autism
And finally, in this task, it could be a good idea to filter out the randomly moving noise dots and just focus on the signal dots, and this could again be limited in autistic children
So if we want to distinguish between these options, we’re going to need another task.
what is the direction integration task?
The other task we can use is a direction integration task, where, rather than having separate sets of signal dots (which I’ve highlighted in red here) and noise dots, here we have the dot directions taken from a single Gaussian distribution on each trial, and the difficulty is manipulated by varying the standard deviation of those directions.
Children are still asked to work out the overall motion direction
what is the second develop direct integration task?
If we use this second direction integration task, we can also apply equivalent noise modelling, which allows us to estimate a child’s sampling ability (i.e., how well they can pool over multiple dots) and internal noise (i.e., how well they can work out the direction of each individual dot direction).
Motion coherence task = integration + segregation
Direction integration task = integration but NO segregation
what is the equivalent noise paradigm?
gives estimates of internal noise and sampling
SD = 0 means all dots moving in the same direction
The idea is that variability in our performance is determined by both external noise – so noise in the stimulus – and noise inherent in our nervous systems – internal noise.
The idea is that we can manipulate the variability in the stimulus (i.e., the external noise) to get an estimate of the internal noise.
We manipulate the stimulus noise in a Gaussian motion task by changing the SD of the distribution from which the dot directions are taken.
black = the normal way that people use the equivalent noise paradigm. You measure a direction discrimination threshold – so what is the smallest difference left/right of vertical that can be detected – at a range of different levels of external noise, or variability in the signal. Here, the SD of dot directions is 0 degrees, so all the dots are moving in the same direction, and there is no external noise. Here the dot directions are taken from a distribution of directions with a large SD, so there is high external noise. We can get an idea of the level of an observer’s internal noise by seeing at what point the thresholds increase. At small levels of external noise, internal noise dominates. However, as we increase external noise, we get to a point where the internal noise inherent in the system is swamped, and so thresholds increase with increasing external noise.
The whole function will be shifted up if individuals sample or average over fewer dots to make their decision. So from the equivalent noise function you can get both estimates of internal noise and sampling.
However, this way of getting an equivalent noise function requires many hundred trials, as each point on here is an individual threshold. Which is just not feasible with children. So we used an efficient method where there are just 2 conditions – here we are looking at the grey line.
In the no noise condition, there is no noise in the stimulus – i.e., the standard deviation of dot directions is 0 deg, and all of the dots move in the same direction. We are then interested in seeing what the finest direction discrimination possible is, from left or right of vertical
In the high noise condition, the mean direction of dots is 45 degrees, and we see how much stimulus noise can be added (so how much we can increase the standard deviation of dot directions) until accurate direction discrimination breaks down.
These two conditions constrain the fit of the equivalent noise function, and from this, we can get estimates of internal noise (so the precision with which individual dot directions can be discriminated) and sampling (how many dots are being averaged over).
what were the results for the equivalent noise paradigm task and what are the implications?
33 cognitively able autistic and 33 typically developing children 6 to 13 years,, matched in age and PIQ (performance IQ)
In the no-noise condition, the absolute direction thresholds were similar between autistic and typically developing children, with no significant group differences.
We were expecting autistic children to have difficulties integrating the motion signals in the high-noise condition, but instead we found that autistic children were able to work out the average direction over a greater range of directional variability in the stimulus than typically developing children – showing enhanced integration ability.
We also found no significant differences in motion coherence thresholds in this study.
When these thresholds were fitted with an equivalent noise function, we found that there were no significant group differences in internal noise, but that the autistic children had significantly increased sampling – suggesting that they are effectively able to average over more dots than typically developing children.
Increased sampling should lead to reduced motion coherence thresholds, so the fact that we don’t see significant enhancements here could be because autistic children are limited in motion coherence tasks by their ability to segregate signal from noise. So while they may be better able to average motion information, they may have difficulties in filtering out the randomly moving noise dots in motion coherence tasks.
summarise people with autism’s performance with motion coherence tasks
▪ Autistic children can average motion over a greater directional range than typically developing children
▪ … consistent with increased sampling (goes against the predictions from the Weak Central Coherence theory)
▪ No benefit in motion coherence task due to reduced segregation of signal-from-noise in autistic children?
▪ n.b., similar suggestions about reduced segregation of signal-from-noise have been made in dyslexia (sometimes “noise exclusion”, e.g., Sperling et al., 2005)
overview dorsal stream vulnerability
Dorsal stream “slower” to develop than ventral stream
This could make dorsal stream functions more ‘vulnerable’ than ventral stream functions to atypical development (e.g., Braddick et al., 2003) – in a range of conditions
Motion coherence - dorsal stream
Form coherence - ventral stream
what dorsal stream vulnerability was found in dyslexia?
the best test is when you have well-matched form and motion tasks
Hansen et al. (2001): significant group differences only for motion coherence
(but some suggestion of group differences in form task too, and only 15 dyslexic participants in the sample)
what dorsal stream vulnerability was found in autism?
Specific elevation in motion coherence thresholds, not form coherence thresholds (Spencer et al., 2000)
But Spencer & O’Brien (2006) - reduced performance in both tasks with better matched stimuli
Spencer et al 2000 – motion task involved locating a target strip with coherent moving dots oscillating in opposite phase to those in surrounding region
Whereas form task involved detecting a circular region of oriented lines segments.
Instead Spencer & O’Brien 2006 used ‘glass patterns’ – dot triplets - which is like where dots are presented on subsequent frames in a motion task. More comparable. Here they found higher thresholds in both form and motion tasks.
