Date: Tuesday Nov 18 at 9am Pacific Time (no longer daylight saving time).
Guest Speaker: Aleksey Bukinich, PhD in Psychology, Lomonosov Moscow State University
Zoom Link: https://zoom.us/j/98564124237
Meeting ID: 985 6412 4237
Passcode: Vygotsky
Summary:
Alexander R. Luria developed a highly useful framework for analyzing the units of mental functions. He integrated ideas from the Vygotskian cultural-historical approach with data from brain lesion studies to formulate his theory of the systemic and dynamic localization of mental functions. Today, this theory is applied to better understand the nature of learning disabilities in children, particularly those primarily related to cognitive dysfunction. This approach enables psychologists, neurologists, psychiatrists, teachers, and parents to understand the causes of learning disabilities and to plan effective interventions.
However, Luria’s legacy includes both strengths and limitations of the neuropsychological approach to psychodiagnostics. It is primarily quantitative and relies heavily on the diagnostician’s education and experience. Current research in this area aims to integrate a more nuanced qualitative approach with principles from Item Response Theory (IRT) – a robust framework considered essential in modern psychodiagnostics.
A proposed five-factor model for child neuropsychological investigation has been developed to integrate these qualitative and quantitative approaches. The five factors are:
- Executive functions
- Auditory-verbal information processing
- Visuospatial information processing
- Hypoactivity and sluggishness
- Hyperactivity and impulsivity
This model incorporates Vygotsky’s concept of syndrome structure, the established structural validity of neuropsychological batteries, and the possibility of comparing children to developmental norms.
Readings (unfortunately, only one is in English, but the key idea is described):
Hello,
Unfortunately I could not listen to the live presentation, so I had to wait until the recording became available. The title was very intruiging for me. Vygotsky-Luria’s approach is disappearing from (neuro)psychology for no scientific reason. Aleksey’s presentation seems only to confirm that it is so. Luria was well aware of factor analysis as used in quantitative data analysis but nevertheless continued revealing “factors” by studying single individuals (Luria, 1969, Higher Cortical Functions in Man (In Russian), p. 78).
During last half of a century an increasing number of substantial arguments have been provided to show that any kind of statistical data analysis cannot, in principle, reveal the structure of psychical processes. Here just one argument, provided by Peter Molenaar, is hopefully sufficient: intraindividual variablity can be revealed by studying inter-individual variability only if the studied process is ergodic. Psychical processes, none of them, is ergodic. Vygotsky and Luria aimed at revealing the structure of different psychical processes by studying individuals, and only individuals, for a reason.
Actually, Aleksey’s results of factor analyses, which may seem to support structural interpretation, are quite problematic. There are three problematic issues that come to my mind. First, the factors are highly correlated – the correlations between factors are substantially higher than many correlations/factor loading of the measures with respective factors (see Table 1 in the paper attached by Aleksey). I would guess that tests for number of factors (Scree plot, Horn’s analysis, whatever) would reveal one or maybe two factors, not more. Of course, it can be said that the model is hierarchical. But this hierarchy is quantitative; Vygotsky-Luria’s theory assumes qualitatively distinct “factors”. Quality is about all-or-none, it has no relation to gradual quantitative differences.
Second problem is with the factor loadings themselves. Most of the factor loadings indicate that shared variance between an item and a latent variable is less than 30%. I have no idea, how to conceptualize the latent variable as a whole, if unshared variance levels are so high, very often more than 70%.
And third, the pattern of relationships between the items and a latent variable is very confusing. Take, for example, the first latent variable (EF). Many questions come to my mind here. For example, why understanding of instructions and number of errors on Go/No-go task are both positively loaded to the factor? Why, in case of Verbal Fluency, productivity in the third test is related to the latent variable negatively but productivity in the first subtest positively? And errors on the second subtest are also related negatively – together with productivity on the third test? In case of the AV factor, total number of verbal errors and the number of exact plant numbers have both negative loadings? … There are several more similar questions – how correct answers and errors load to the latent variable in the same way? In addition, how to interpret loadings of the left- and right-hemisphere strategies to the VS factor? – Both are relatively high and positive. What would be the essence of the VS factor where qualitatively opposite strategies load to the factor in the same way? And this together with similar positive loading of the right hemisphere errors? So, again, what is the essence of the distinguished factors with such factor loading patterns?
Altogether, so-called mixed-methods approach, used by Aleksey, seems to add confusion rather than clarity to qualitative theory developed by Vygotsky and Luria. The founders of statistical data analysis (e.g., Pearson, Thurstone) knew very well that statistical data analysis does not, and can not, reveal anything about the nature of whatever is studied; it is just a way to deal with complex data. Vygotsky and Luria both knew it also. Trying to force their qualitative theory into quantitative approach seems only to create confusion, not deeper understanding.
With best wishes
Aaro
(Toomela)
From Francine,
Aaro’s concerns about what whether factor analysis can actually tell us anything useful about cognitive functioning are worth further consideration. This does not require a background in statistical analysis, although I appreciate that members of our group have that expertise.
Since correlations can only be calculated for multiple subjects, factor analysis of multiple correlations reveals nothing about an individual’s functioning. If I understand what Aleksey said – a government oversight committee wanted statistical validation of neuropsychology assessments. But another way of providing statistical validation would be a study of outcome assessments, does a qualitative neuropsychology assessment lead to improved functioning of individuals (let’s say on the autism spectrum) compared to quantitative assessment techniques.
Let’s talk more about all this.
In regard to creativity, what has the factor analysis identification of divergent thinking contributed to our understanding of creativity, to our ability to function more creativity?
From Tanya Akhutina:
Dear Aaron!
Thank you for your letter and for your serious approach to both the report and the fate of Vygotsky-Luria neuropsychology. Your letter contains specific questions and a general statement. Since Alexei Bukinich’s report sets out our common view on the need to develop a quantitative approach in neuropsychology (and I am one of the initiators of this approach), your questions are also questions for me.
Your specific questions are valid, and we are ready to answer some of them today, while others require further consideration, and I admit that we cannot answer some of them.
But I and we categorically disagree that a quantitative approach is a death sentence for the qualitative neuropsychology of Vygotsky and Luria. You write: “Quality is an ‘all or nothing’ principle; it has nothing to do with gradual quantitative differences.” Yes and no. Here we enter the realm of dialectics.
Quantitative assessment of qualitative differences is an everyday task of science. Everything is known in comparison. And this is a general law for both science and language. Any comparison implies equalization. When we compare different objects, we single out one common property, keeping other properties in mind. “This is a triangle, this is a quadrangle.” We say this, defining the qualitative uniqueness of objects and drawing attention to one common property—the presence of angles. Can the qualitative uniqueness of objects be reduced to this one property? Of course not, but have we captured something essential in this uniqueness? More often than not, yes. Science is getting closer to fully grasping qualitative uniqueness, but, as in the case of squaring the circle (inscribing polygons with an increasing number of angles into a circle), this infinite approximation does not eliminate the gap. The need for a qualitative leap always remains.
I think that in quantitative studies of qualitative differences, we must not forget about the problem of squaring the circle, but we cannot abandon science because of this. Every word is only a hint, but we do not abandon language because of this.
Tanya Akhutina
Dear Tanya,
Thank you for your response. I think differences between Vygotsky-Luria’s kind of qualitative approach and quantitative approach are much deeper, in fact, incompatible. I am going through your arguments. First, you say that “Everything is known in comparison.” I fully agree. But differences can be qualitative or quantitative. Next, “Any comparison implies equalization. When we compare different objects, we single out one common property, keeping other properties in mind.” I understand it as a question of generalization in the situation where everything has unique characteristics. For example, all things in the universe have environments, the worlds surrounding them, which are in some respect unique. A particular object cannot be in the environment of itself even though it can, and is, in the environment of everything else. Thus, when we compare different objects, we must leave out of the comparison unique aspects of all objects involved and distinguish characteristics that are common to all. All mammals are in many characteristics unique and yet all mammals have a set of identical characteristics. Here is also an answer to your next question: “Can the qualitative uniqueness of objects be reduced to [this] one property?” Obviously not, everything has some absolutely unique characteristics. But objects with one common quality, or a certain set of qualities (e.g., all higher psychological functions, HPFs, are mediated by symbols, by linguistic signs), are all similar in that particular quality. Sometimes, like in case of HPFs, many important qualitative consequences follow. For example, certain kinds of psychical processes are impossible without HPFs. This is not qualitative uniqueness, it is qualitative similarity, which distinguishes a certain type of objects with that quality from everything else. Vygotsky-Luria’s science, as I understand it, is about discovering qualities that distinguish psychical processes one from another.
Quantitative approach is useless here. Your example of squaring the circle is interesting here. Lindermann-Weierstrass theorem explains how it is mathematically impossible. Furthermore, squaring a circle, that is, constructing a square with the area of a given circle, is purely quantitative problem. Thus, the question is how to find a certain similarity (area) of qualitatively different geometrical objects (square and circle). And it turns out to be impossible, at least in that particular property.
Vygotsky and Luria (among several other psychologist of their time) aimed at revealing the structure of psyche. For that, they searched answers for three questions: what is the studied whole? (e.g., HPF); what are the parts of that whole? (e.g., words/linguistic signs and other psychical processes, such as memory and attention); and in which particular relationships these parts must be in order to make the studied whole? Answers to all these questions must be qualitative. Certainly, some quantitative aspect is there too. There can be no structure composed from one indivisible component; there must be at least two. But such quantitative aspects cannot provide any understanding of what is studied.
Furthermore, quantitative data analysis cannot, in principle, reveal the structure of psyche (or anything else). One common problem, as Peter Molenaar showed, is that analysis of inter-individual variation can be generalized to intra-individual variation of single cases only if the modelled process is ergodic. Psyche/mind is a property of an individual. Psychical processes are not ergodic. Therefore, all quantitative analyses of individual differences are useless, actually misleading, if the aim is to understand psyche, that is, an individual.
It may seem that quantitative analysis of intraindividual variation may still be an option. But it is not for quite many different reasons (I have discussed several of such problems in some papers, e.g. https://doi.org/10.1007/s12124-008-9059-6 and https://doi.org/10.3389/fpsyg.2010.00029). As an example, results of quantitative analyses in your paper also demonstrate that Luria’s qualitative distinctions disappear in quantitative analyses. Even if you would find “explanations” to such discrepancies, these explanations would be ad hoc. If you would use other sample or other tests to assess qualitatively the same constructs, another, different pattern of relationships would follow where these ad hoc explanations would not apply. Furthermore, the results of analysis of covariations between the same set of variables would be different depending on which specific method of analysis is used. There is also not substantial criterion to determine which specific analysis would be appropriate for a specific set of variables or questions about the covariations between the variables. Vygotsky did not need quantitative methods, Luria did not need them, but their theories are reliable and valid. Quantitative methodology is a death sentence for qualitative neuropsychology.
With best wishes
Aaro
This looks to be very interesting