On 2016 Apr 20, Lydia Maniatis commented:

The authors of this study have constructed an experimental setup that has generated data that are “consistent with a stochastic process that degrades memory for each individual item and plays out independently across them.” On the basis of this, they conclude that “visual working memory is stochastic both at the level of a memory's content and at the level of its quality.”

A few objections to this assertion are in order. I want to preface them by noting that it has long been understood that it is always possible to generate data “consistent” with any point of view: “A nice adaptation of conditions will make almost any hypothesis agree with the phenomena” (J. Black (1803, Lectures on the Elements of Chemistry, quoted by Popper in Popper Selections, Miller, Ed, 1985). In order for “consistency” to carry weight, investigators must have really tried to look for “inconsistency” or falsification, wherever it seems most likely to be found.

As it happens, there is a preponderance of evidence that memory in general and visual memory in particular is not stochastic either in its content or its quality. With respect to visual memory, the organizing principles that influence the percept itself also influence the memory of those percepts, as is shown, for example, by Goldmeier (1981, The Memory Trace, its Formation and its Fate). Other investigators who have shown that memory changes are not random include Bartlett and Loftus.

If the authors want to challenge the legitimacy of the view that “degradation” and random change is not the proper account of memory, then they should do it. It would be fine if they validly challenged the results of studies that falsify their position. They could, for example, claim the methods of the falsifying experiments were flawed. What is not OK, however, is for scientists to ignore contradictory evidence, and set up experiments that produces “consistency” with their preferred view.

Significantly, the task used by Fougnie, Suchow and Alvarez' (2012) involves a structure-less feature – color - thus ensuring that the known effects on memory of structural principles will not interfere with their preferred outcome. As they note in their introduction, this is a task that is commonly used for such studies. Unfortunately, as here, many studies today rationalize their methodological choices not on theoretical grounds, but on the grounds that they are widespread, popular, common, even "traditional" (!). As here, a method may have become normalised for the wrong reason, i.e. because it doesn't complicate the perpetuation of preferred paradigms uncritically accepted.

In my discussion above I have not distinguished between “working memory” and other types of memory. The authors, while using the term “working memory” have not provided a working definition, i.e. a description of how they circumscribe the term and distinguish it from other types of memory. Once they do this, they might try and check the literature for examples that falsify their account, and construct experiments that, on the basis of what has already demonstrated, make a credible effort to falsify it, rather than simply to “confirm” it.

On 2016 Apr 20, Lydia Maniatis commented:

The authors of this study have constructed an experimental setup that has generated data that are “consistent with a stochastic process that degrades memory for each individual item and plays out independently across them.” On the basis of this, they conclude that “visual working memory is stochastic both at the level of a memory's content and at the level of its quality.”

A few objections to this assertion are in order. I want to preface them by noting that it has long been understood that it is always possible to generate data “consistent” with any point of view: “A nice adaptation of conditions will make almost any hypothesis agree with the phenomena” (J. Black (1803, Lectures on the Elements of Chemistry, quoted by Popper in Popper Selections, Miller, Ed, 1985). In order for “consistency” to carry weight, investigators must have really tried to look for “inconsistency” or falsification, wherever it seems most likely to be found.

As it happens, there is a preponderance of evidence that memory in general and visual memory in particular is not stochastic either in its content or its quality. With respect to visual memory, the organizing principles that influence the percept itself also influence the memory of those percepts, as is shown, for example, by Goldmeier (1981, The Memory Trace, its Formation and its Fate). Other investigators who have shown that memory changes are not random include Bartlett and Loftus.

If the authors want to challenge the legitimacy of the view that “degradation” and random change is not the proper account of memory, then they should do it. It would be fine if they validly challenged the results of studies that falsify their position. They could, for example, claim the methods of the falsifying experiments were flawed. What is not OK, however, is for scientists to ignore contradictory evidence, and set up experiments that produces “consistency” with their preferred view.

Significantly, the task used by Fougnie, Suchow and Alvarez' (2012) involves a structure-less feature – color - thus ensuring that the known effects on memory of structural principles will not interfere with their preferred outcome. As they note in their introduction, this is a task that is commonly used for such studies. Unfortunately, as here, many studies today rationalize their methodological choices not on theoretical grounds, but on the grounds that they are widespread, popular, common, even "traditional" (!). As here, a method may have become normalised for the wrong reason, i.e. because it doesn't complicate the perpetuation of preferred paradigms uncritically accepted.

In my discussion above I have not distinguished between “working memory” and other types of memory. The authors, while using the term “working memory” have not provided a working definition, i.e. a description of how they circumscribe the term and distinguish it from other types of memory. Once they do this, they might try and check the literature for examples that falsify their account, and construct experiments that, on the basis of what has already demonstrated, make a credible effort to falsify it, rather than simply to “confirm” it.