8790653042.pptx
Okay, Hello, I am the next presenter, as you know, Hunjae Lee.
The article that I chose is “Texture density adaptation and the perceived numerosity and distribution of texture” by Frank Durgin, 1995
It looks weird because the former presenter Sungmin already mentioned this author, Durgin with his 2008 article. And I show you his study published more than 10 years ago. But this is good one, a work of great labor. so pages are more than 20. He thoroughly review previous work. in this time I cannot present whole of work, I will just try not to miss the essence of this study. please listen me with generous mind but be awake.
/0
Introduction.
In daily life when we see the complex texture, like this clothes. detail informations are lost. only abstract summary informations are available. then what’s the contents? luminance? color? spatial frequency?
Here, in this study, Durgin concentrate on the scatter dot.
What do we see when we look at scatter-dot texture?
And his answers are numerosity and cluster. and He reveal that the perceptions of them interact each other.
/
Now you already know this method of test perceived numerosity. So I skip this.
Demo.. skip skip..
/
From his intensive review of previous study about Numerosity, density and cluster. I picked up just two that is directly relates to his prediction.
First, Occupancy model by Allik & Tuulmets, 1991. It is kind of theory that consider numerosity perception as magnitude system. 읽는다.
This occupancy model has only one parameter radius R, but explain the perceived numerosity of people very well. But this model is tested under 20, 40 dots. relatively small numerosity
And second one, Barlow suggested that there is distinction between numerosity judgments for low numerosities (but more than subitizing, counting range) like 40 and discriminations based on texture or density for higher numerosities like hundreds of dots.
So Durgin made his prediction. For textures with numerous elements, in other words high density, adaptation reduces the apparent number of texture elements but does not distort their apparent cluster. But For low-density textures, nu-merosity suffers only slight distortion, but a large change is found in the apparent cluster of texture elements.
This picture show the stimuli very well, that is manipulated in this study. 읽는다.
And Here is experiment 1 and its data. In experiment 1, They use five staircase to acquire PSE(Point of subjective equivalence) with this five number. And interdot distance is fixed. One group of participants take the numerosity comparison and the another group report perceived density. As you can see the graph.
The results confirm that the perception of numerosity and the perception of density are affected differently by adaptation. Adaptation produces a strong and consistent distortion of perceived density at all levels of numerosity. On the other hand, the relative distortion of perceived numerosity increases significantly with test numerosity increases. This implies that texture density information has an increasing influence on perceived numerosity as numerosity grows
//
now, Main point of this study. Experiemnt 2. they manipulate minimum interdot distance in the staircase of 3 number, 20 80 320 at adapted field and 2 numerosity at unadapted field. PSE is acquired from the first phase of experiment. so total 6 staircase are conducted to get minimum interdot distance which make PSE at density comparison task.
MinIDmax can be obtained as a square root of area divided by number of dots. So this MinID divided by MinIDmax is kind of relative ratio regardless of big change in numerosity. And graph clearly the opposite pattern for perceived cluster. There is no evidence of a systematic distortion of perceived cluster at the highest numerosity but in the range of small numbers, the greatest distortion of perceived cluster occurs.
After the discussion with his experiments, Durgin want to modify and extend the occupancy model that can fit well with his new findings. So we go back to occupancy model, Here we can see the problem of occupancy model very easily.
Model predicts that the right field should appear more numerous (occupied area is outlined area). It is because the occupancy model was derived with low numerosities, and is not sensitive to the density information in these textures.
The problem originated from the radius R. Because Allik assumed the radius R with their behaviral data that is most suitable for explaining data, and it fit quite well various data in their study. and they didn’t say not much about R, what is the factor to determine the size of R. If the R is small then occupancy model still fits well.
//
So Durgin have simulated occupancy model with various R size. Very beautiful graph here. Allik and Tuulmet use the size 20 of R in their original occupancy model study. considering x axis, If the dots are more than 100 the perceived numerosity is severly impaired with this occupancy model. one hundred is a advice of professor chong.
This graph show the cluster effect on perceived numerosity with various R. here, they manipulated cluster with ratio of regularity that I mentioned before and number of dots and size of radius. Y axis is the ratio of the change of occupancy index which means that the size of filled area. when R is set to 20 it can reflect the change of cluster at low numerosity like 20, 40 but it cannot at high numerosity 160, 320 , the ratio 1. Conversely if the size of R is too small like 5 and low numerosity It is also not much affected by cluster, right?
Although judgments of texture density itself are uniformly and strongly distorted at all levels of density, the effects of adaptation on numerosity and clustering
judgments appear to be range dependent and reciprocal:
//
Conclusion and Discussion. 읽는다.
//
My additional article is numerical cognition, reading numbers from the brain. by Kadosh and Walsh. It is very recent one but I think Chong already knew this one also. Because this article is kind of sum up and some opinion about this article Eger et al… 읽는다.. at the same issue of Current Biology. So I failed to serve new one to Professor. And not that good relationship with my presentation. I am sorry about this.
Anyway, This is overview of their experimental design.
Sample number stimulus and, after a random delay period, a match stimulus appeared that differed in numerical magnitude by 50% and required a numerical smaller versus larger judgment.
In experiment 1, they use two different nonsymbolic stimulus lists. to control overall luminance or dot size between numerosities. Number 4, 8, 16, 32 were used.
In experiment 2, they use format change between sample
and match occurred in 50% of the trials with these set of stimuli
They used multivariate pattern recognition with fMRI data from experiments
Parietal activation patterns for individual numerosities could be discriminated and generalized across changes in stumuli.
Distinct patterns were evoked by symbolic and nonsymbolic number formats,
and individual digits were less accurately decoded than numbers of dots. Interestingly, the numerosity of dot sets could be predicted above chance from the brain activation patterns evoked by digits, but not vice versa.
———-
Finally, Discussions
So as Burr argues, Durgin is on the corner? with this neurophysiological data nowadays? numerosity perception has direct representation in our brain rather than concerning complex processing with texture density or other visual property ?
I don’t know well. but my feeling is just pro for Durgin. He made long effortful paper… Um.., And I think Burr and many researchers skipped over many low level processing to percieve numerosity. I feel like there should be missing link… But I don’t know.