Baddeley, Alan. (2000). The episodic buffer: a new component of working memory?

Baddeley, Alan. (2000). The episodic buffer: a new component of working memory? Trends in Cognitive Sciences, 4(11), 417-423.

The concept of working memory by Baddeley and Hitch:

1. It abandoned the concept of a unitary store in favour of a multi-component system
2. It emphasized the function of such a system in complex cognition, rather than memory perse.

Anatomical affiliation

• Phonological loop (BA40, BA44)
• Visuospatial sketchpad (BA6, 19, 40, 47)
• Central executive (... frontal lobe areas)

Also... the loop... is it designed for language acquisition while the sketchpad, visual semantics?

My thoughts on the notion of language... 

1. What is a language? 
2. What is a vocabulary?  
3. What is grammar?  
4. What is syntax?...

(Why entertaining this/these?)

Back to the paper, where does the component of Episodic Buffer come from?

Problems for the current model (as of 2000)

1.1. The phonological loop: limits and limitations

The phonological contains both a temporary storage and the articulatory rehearsal process. Without the articulatory rehearsal process, traces of the auditory memory decays within a few seconds.

This model use useful in explaining the following phenomena:

• The phonological similarity effect: Items more similar in sounds are harder to remember although visual or semantic similarities have no impact.
• The word-length effects: Because longer words takes longer time to rehearse and reproduce, the longer time requires to do these makes it easier for traces of memory to deteriorate.  
• The effect of articulatory suppression: When people are prevented from rehearsing items to be remembered, their recall performance decline and the word-length effect disappears.
• Transfer of information between codes: Articulatory suppression removes the phonological similarity effect for visually presented items but not for auditory ones.

(Come to think about it... there is something interesting about the transfer of information between codes... In times when I was very drugged out though not yet to the state of from 2-3 words, I had problems reading while comprehending the texts.  Yet, I found out that I had no problem listening to the same texts in auditory format.  So I had the computer reading the texts to me.  Also, sometimes, when typing, I had to type things out without looking at the screen and that's how thoughts can flow out... Feels as if these have something to do with the notion of transferring information between codes.)

• Neuropsychological evidence: What is essential for rehearsal is the central rehearsal code (i.e., speech motor code in dyspraxic patients) rather than its over operation.

There are phenomena that doesn't fit the model

• 1. Recall on visual representation of sequences of numbers do not suffer traumatically with the concurrent articulatory suppression task (i.e., pronounce words such as "the").  Generally, the digital span only drop from 7 to 5.
• 2. Patients with impaired phonological memory can recall about 4 digits when stimuli were presented visually.

What about visuospatial sketchpad? Theoretically, it's for the storing of patterns... not suited for digits.

Visual similarity effect?  In Logie et al., the authors found visual similarity effect except the effects are small and not limited to the articulatory suppression condition. (Under the assumption that, if you can count on the phonological loop, visual similarity or not should be an issue.)  Since the visual similarity effect is present during non-suppressed condition, it means the performance is not all based on the phonological loop... therefore, the info from these two slave systems might be combined somewhere.

1.2. Prose recall

With unrelated words, people can remember up to 5-6 words before starting to make errors.  However, in the context of a meaningful sentence, the span can be as high as 16... a manifestation of Miller's "chunking." Based on knowledge in LTM, the sentence is broken down to smaller chunks with the capacity set by the number of chunk as opposed to words.  But where are the chunks stored?  In phonological loop, LTM, or a different storage space?

The phonological loop hypothesis doesn't fly in the case of patient "PV." Patient "PV" had the loop capacity of 1 word and intact LTM has a sentence span of five.

This reminds me of the recurrent eras of from 2-3 words.  I was so drugged out that I couldn't read and couldn't understand people if they spoke in longer sentences because, when they get to the end of the sentence, I would have forgotten what they said before.  It was the drug effect.  What I eventually had to learn to do is... when reading, because my digit span went as low as 1 (like... what was the last word I read? Forgotten already), instead of trying to comprehend the sentences (extremely high tech), if not simply terms, I identified key words or chunks of words and deliberately tried to find them from LTM.  Sure, it took an awful lot of time to finish a sentence or two, and what was read was already forgotten, the way I saw it... at least some part of LTM got activated... path strengthened... and hopefully, it would make it easier for me to read the same texts again when, if, one day, I could grow out of from 2-3 words.  Did it work and did anything got retained?  I don't know.  One thing I do know is I did get to read something with a digit span of from 1, 2, 3 words.  

Also, while severely amnesic patients are uniformly bad at recall after a 20-minute delay, their immediate recall of a prose paragraph containing 15-20 idea units could be relatively normal.  The issues.... the amount of items far exceed the capacity of phonological loop with the recalling process might overwrite what's in it.   What about visuospatial sketchpad? Theoretically, it's unsuitable for serial recall and there is no evidence that only imageable sequences can be recalled.  What about the central executive?  Theoretically, it has no storage capacity.

It takes both the retrieval of knowledge in LTM and the manipulation on what gets retrieved to envision an ice-hockey-playing elephant. Sure the process can take place in LTM.  Yet, there is no evidence claiming patients with normal intelligence and grossly impaired LTM cannot get the elephant envisioned.

Also, Tulving had seen a densely amnesic patient to play a good game of bridge and win the game.  He apparently could keep track of what was going on and which cards had been played... a task beyond the assumed capacity of the two slave systems.

1.3. The problem of rehearsal

Think about the notion of rehearsal in phonological loop... equivalent to subvocal rehearsal adults often do when trying to memorize something.  The subvocal rehearsal is more advantageous than rehearsal in other modalities in two ways: First, people can generate the items by speaking them out... making it easier to be in a covert format.  Second, digits and words have intrinsic lexical representation, making it easier to repair the deterioration of memory during the rehearsal process.

Yet, how could the rehearsal be done in the visuospatial sketchpad or the other back-up store?

1.4. Consciousness and the binding problem

Existing studies showed the role of consciousness in information processing.

In this section, this author tried to explain to the readers that the model is lacking a component where complex info from different channels in all modalities (e.g., slaves systems, LTM) could be stored, manipulated, integrated, and controlled by the central executive.... a solution to the binding issue.

The episodic buffer

The episodic buffer is episodic because "it holds episodes whereby information is integrated across space and potentially extended across time, which makes it resembles Tulving's notion of episodic memory.  However, it is temporary and can feed and receive information from LTM. It's a component that interface with multiple systems and codes.  It's capacity is assumed to be limited due to the high demands in the tasks involved.  Also, it can be accessed "by the central executive through the medium of conscious awareness" and CE can therefore decided what to do about the information... attending to it or not, creating new cognitive representation or not, etc.

There is something interesting about the notion of episodic buffer and it's relation with LTM.  Reminds me of the strange psychotic experience of the reenactment of past psychotic episodes.  How did it happen? Either the central executive went out of whack and decided to do some recherche du tempt perdu through replaying some past psychotic episode.  Or, the out-of-whack potentially automatic and effortless process of knowledge retrieval decided to get the whole set of episodic memory replayed.  Regardless, the entire muti-modal episode need to be placed somewhere... a perfect place to be... episodic buffer.  At the same time, though I had no control over the retrieval and reenactment of the episode in my head, I was capable of trying my best to keep it running in the back of my head... albeit the inevitable bodily sensation.

2.1. How is the buffer implemented biologically?

fMRI studies showed greater right frontal activation for integrated info while the activation of unintegrated retention were more in the posterior areas (verbal and spatial WM).

2.2. So what's new?

Episodic buffer as a new component or a fractionation of Central excecutive.

The revised model:

• WM as multi-component
• WM is not simply the activated portion of LTM
• Reject the view that "the slave systems merely represent activations within the process of visual and verbal perception and production." (Bye bye to dual channel)
• WM is considered a seperate system from LTM and perceptual and motor function because it has it's own "dedicated storage process.

Some outstanding issues

"conscious monitoring of the evidence supporting an apparent memory plays a crucial role in separating accurate recall from false memory, confabulation and delusion."

Not quite sure where the notion of delusion comes from.. since I have entertained this notion else where... in short, a multi-dimensional concept.