Disclaimer: English Kinda Thing

The sole purpose of the "English Kinda Thing" is to document my attempts to correct my own mistakes in standard English usage and to share the resources I find. In no way do I attempt to teach nobody English through these blurbs--just as I intend not to teach nobody to be a neurotic and psychotic handicap in Ratology Reloaded or Down with Meds! :-)

Sunday, June 23, 2013

Baddeley, Allen, & Hitch (2011) Binding in visual working memory: The role of the episodic buffer

The notion of binding is an interesting one... reminding of the paired-associations I have entertained else where...  (too many elsewhere in the dungeon of limited my words that I can keep track not where them words are.  lol 8-X)

On a lousy weather afternoon... done taking notes on this paper...  Lightening and thunder running wild on the outside paired with pouring rain when I plowed away jotting down notes on the paper "Binding in visual working memory: The role of the episodic buffer". Like splitting on the outside while binding on the inside? lol

For psychotics, one skill we have to learn is to unbind... at the feature level, object level, or in the episodic buffer or not.  The good thing, according to the following, is... you can do something to interfere... Let's jam up the limited capacity with alternative processes... imposing excessive workload to disrupt the fragile bound object!  Will it work.  Donno.

Baddeley, Alan D., Allen, Richard J., & Hitch, Graham J. (2011). Binding in visual working memory: The role of the episodic buffer. Neuropsychologia, 49(6), 1393-1400.

What is working memory for Baddeley, Allen, & Hitch?
"We use it principally to refer to a broad framework of interacting processes that involve the temporary storage and manipulation of information in the service of performing complex cognitive activities."

  1. The two slave systems: Phonological loop and visuospatial sketchpad
  2. The central executive (with no capacity for information storage).
  3. The episodic buffer: A storage "accessible through conscious awareness and provides the milti-feature binding mechanism-- controlled by the central executive.  (Somewhere in time but I don't remember when... I remember reading that episodic buffer is sort of like a spin-off of the central executive.) 
The model as of Baddeley (2000)

1. Exploring the episodic buffer

As seen in the above diagram for the model as of Baddeley (2000), originally, it was assumed that everything bound by the buffer has to go through the central executive.  Under this assumption, it's reasonable to assume that the disruption of CE function should disrupt the binding of features into an integrated presentation... such as that ice-hockey playing elephant.

Research paradigm: using dual task methodology to study the binding of features in visually or verbally.

(As I was reading about that ice-hockey playing elephant yet another time the other day, this man on the street was talking about the elephant he was trying to sell.  When stealing a smoke outside of the cafe, I saw these two bronze elephants... and made an interesting observation...  It's more difficult to look at an elephant and try to envision it playing ice-hockey... the binding of the image doesn't go as smooth and stable as binding the image up... frame by frame not not... by retrieving the features directly through LTM. Don't know how common it is...  It was quite funny though... like the LTM-based elephant was playing ice-hockey on top of leisurely throwing out triple toes for the sake of it while the one based on both LTM and external visual stimuli could barely keep the gloves to its hand.  Funny enough... given that I am psychotic, my head should be really skilled in this binding thing since it could bind the most irrelevant stimuli... shared by the others or not... It might speak of the unmentioned power of LTM or the interfering effect of real visual input... unless the central executive was simply in a mood of chilling and didn't want to work too hard? 8-O lol 8-X)

2. Binding in visual working memory

  1. Limited storage capacity: Automatic biding of features into objects with a storage capacity of 4 objects (regardless the features for the objects).
  2. Limited capacity for attention: Its attentionally demanding to combine features into objects and maintain them in the visual working memory (Wheeler and Treisman, 2002) (You see Treisman... you can guess that it's about attention.)
  3. The third possibility: The initial binding of features into object is automatic but the maintenance takes effects. 
(One problem of disordered thinking is that I can't stop my head from binding things together... faulty automaticity at fault... not me...)

The rationale behind the studies done to address the third possibility:  The participants had to carry out simple or complex verbal task (attention load) while performing feature and binding task (encoding phase).  Since the concurrent task is verbal in nature, shall there be interference effect, it's gotta be the issue of the workload on the central executive rather than the visuo-spatial processing.

So they say, a picture speaks a thousand words... the following figure shows you all the studies done to test the third possibility.  While a, b, c, d, and e represents how the color and shape features are presented, f and g are the recognition probe (i.e., did this appear in the previous presentation?)

Study result of presentation a: 

Counting backwards 3 digits interfered greatly with performance accuracy, which indicate the role of CE.  However, the effect was no larger for binding than for features.  Findings of this study was replicated in subsequent experiments utilizing different concurrent tasks.  These findings seemed to suggest that the binding of features into objects in visuospatial sketchpad is automatic and does not require [much] attentional resources.

The issue... could the binding of shape and color something too "basic, rapid and automatic" to impose cognitive load?

3. Binding across space, time and modality

In three sets of studies, the features were presented in separated fashion...

  • c: separate presentation of features
  • d: separate presentation of features in time

Results: Performance in c and d was above chance level although not as accurate as in the unitized presentation (a).

  • e. separate presentation of features in modality

Results: Performance in e was as accurate as in the unitized presentation (a) and was no more disrupted by concurrent backward counting task.

Thus, "objects requiring the binding of features across space, time, or modality are no more disrupted by an attentionally demanding task than are those for which the features are unitized at presentation." Yet, such findings were inconsistent with the hypothesis that episodic buffer is at the mercy of the CE.

Love this part...

"Our finding that disrupting Gestalt principles does not engage executive processes in binding might seem to imply that the perceptual system binds all features automatically.  However, if this were the case it would surely result in perceptual chaos, potentially combining totally unrelated features to create illusory objects."

Ya... this is life with psychosis.  Like... my head's ability to bind things together is my demise.  Therefore, the pursuit to find ways to unbind.  Since normal people don't do it, it shall mean that some part of the binding between the psychotics and the normal is different... CE related or not.

4. How are bindings maintained?

Though there seems to be automaticity in binding, results of some studies found the retention of found objects to be more fragile than those of the features.

b: sequential presentation as opposed to presenting stimuli in an array

As you could see easily... recency effect... while performance on bound object suffered more than features in earlier items.  Also, when comparing sequential vs. simultaneous presentation, memory on bound object was significantly worse than feature objects in sequential presentation.  In terms of performance under the concurrent task of counting backwards in decrements of two, there was no differences between the sequential and simultaneous presentation conditions; thus, the fragility in bound object is independent of attentional load.

The unanswered question: What contributes to the differential findings in the sequential and simultaneous presentations?

What is forgetting? A reflection of limited storage or an over-write process when participants intend to register new items?

Suffix paradigm: Presenting a suffix the participants have to ignore after an array of stimuli.  If bound objects are more fragile or susceptible to overwriting, it should be more disrupted by the suffix.  The "implausible" suffix does not contain features drawn from the target set while the "plausible" suffix doe.

Implausible suffix: Small but statistically significant decrements though no major differences between object and feature memory.  It didn't matter whether the suffix changed from trial to trial or not.
Plausible suffix: As you should be able to see in the figure below, the plausible suffix caused significantly more disruption on the memory of bound object than that of features. In terms of the level of decrements, smaller decrements similar to those observed in the implausible suffix condition were observed.

The above experiments applied the probe recognition paradigm (e.g., 2f, 2g).

Cured recall experiments were conducted to replicate findings of the probe recognition paradigm.  It was found that plausible, semi-plausible (i.e., stimuli consisting features from one dimension but not the other) had equivalent effect on cued recall... both larger than the impact of the implausible suffix.

What are the lessons learned?

Disruption has nothing to do with the intent to encode the items or not but has to do with the nature of the item.  The findings also reflect two processes:

  • Filtering mechanism: To-be-ignored items are required to be filtered out before entering the WM.  Given the equivalent impacts of plausible and semi-plausible suffixes, it suggests that the filtering operates at the feature level. "as stimulus-based filtering draws on knowledge of task demands and what has been previously encountered, this process reflects one way in which the products of prior experience can impinge on perceptual input before information is encoded into working memory."
  • A process of overwriting:  It's assumed that anything passing the filter is automatically combined as an object.  This level is particularly susceptible to overwrite and the loss of representation... a way to explain why bound objects suffers more than features. At the same time, the assessment through recognition probe or cued recall intends to assess memory of items at the object level. With features, there is a greater redundancy since they appear both at the feature and object level.

Such findings provides supports for the assumption of the displacement in a limited capacity store rather than  feature-based interference (where fully plausible suffix should be more distruptive than semi-plausible suffix).

"Visual stimuli are at least initially represented at both the level of constituent features and of objects.... with filtering proceeding on the basis of features and overwriting particularly impacting on object representation."

How does the above have anything to do with the Working Memory model?

Features are bound into objects in the visuospatial sketchpad before being fed into the episodic buffer.  It's in episodic buffer when objects were conscious retained and manipulated as per the command of CE.

5. Does binding depend on the hippocampus?

Findings of a lot of studies seem to show that binding seemed to involve hippocampus.

Then, along came Jon who had reduction of about 50% of both left and right hippocampal regions while, in comparison to the performance of 6 undergraduate students from the University of work, his performance in a range of visual binding tasks as described earlier didn't seem to differ all that much from the college students... In reality... better at times...

This other lady with a left medial splenoid ridge meningioma removed was found to have a very marked deficit in the short-term binding of color and shape although the intact ability in long-term visual or short-term verbal binding.

6. Binding and the episodic buffer

What about verbal binding?

"As with the visuo-spatial binding experiments just described, we found no interaction between concurrent load and binding. The binding of words into chunks therefore appears to reflect automatic processes, this time clearly involving long-term memory. It does not appear to depend on the central executive."

Automaticity or at least semi-automatic.

Following is the revised model of working memory as of 2011.

To end this note... in the authors' own words...

"Our results do however place constraints on such models, requiring an explanation of (1) the lack of an effect of attentional load on the capacity to bind the features of shape and color, (2) the fragility of such binding, (3) the capacity to filter out “implausible” but not “plausible” suffixes and (4) the mechanism by which plausible suffixes disrupt the retention of bound objects."

No comments:

Post a Comment