[Humanist] 27.1013 varying the cognitive span

Humanist Discussion Group willard.mccarty at mccarty.org.uk
Fri May 2 07:19:26 CEST 2014

                 Humanist Discussion Group, Vol. 27, No. 1013.
            Department of Digital Humanities, King's College London
                Submit to: humanist at lists.digitalhumanities.org

        Date: Thu, 01 May 2014 12:21:29 +0100
        From: Willard McCarty <willard.mccarty at mccarty.org.uk>
        Subject: Varying the cognitive span

To my mind historian and philosopher of science David Gooding 
(1947-2009) wrote several of the most intellectually rich and powerful 
explorations of his subject, the experimental sciences. I just finished a 
slow read of one of these, his "Varying the cognitive span", in Hans 
Radder, ed., The Philosophy of Scientific Experimentation (Pittsburgh, 
2003). I'm about to spring a small bit of it on you, as temptation to read 
the whole of it, because I think it's among the best I know to help us 
explore our own subject. I would hope for some discussion to be 
provoked by the following. 

Perhaps I should recommend a mutatis mutandis but I really think 
that unnecessary. If discussion of the sciences causes you to break 
out in spots or foam at the mouth, then please consider this a health 
warning :-).

He begins:

> Increased dependence on instruments to access primary, abstracted
> features of the physical world marks an important change. Although
> this change clearly displaces ordinary human modes of perception and
> cognition, it does not eliminate them. We can characterize this
> schematically as follows. Aspects of the world are selectively
> redescribed to make them amenable to manipulation according to rules
> that in many cases are now implemented in machines. The effect is
> that certain human modes of cognition and certain skills apparently
> cease to be relevant. They are replaced, to a greater or lesser
> extent, by the very limited modes of cognition of a machine.

He cites the example of medical diagnostics, whose early automation 
proved embarrassing to people surprised by how much a statistically 
based expert system could in fact do. He then picks up clinician Marsden 
Blois' term "cognitive span":

> At its widest, our cognitive capacities must confront the world as
> experienced from moment to moment. At its narrowest, only a few
> highly specialized, skilled, and context-specific capacities are
> needed. A typical process of diagnosis would begin with a wide "span"
> or range invoked when a patient first enters the consulting room:
> there is a preliminary conversation, reading of body language, taking
> a history, conducting a physical examination, and so on.... Together
> these make up the human ability of making "clinical judgements" about
> possible causes of the symptoms identified... Later in the process,
> results of laboratory tests or X rays reduce the number of possible
> diagnoses to a small set of most probable conditions, for which a set
> of interventions may be specified.

Here's where the expert system becomes effective.

Gooding points out that the narrow end of the reductive "funnel", as 
Blois called it, couldn't happen without the wide end, where human 
cognition operates. (We have only promises that computation will extend 
back to the initial phase of diagnosis.) But the argument isn't 
finished. He goes on:

> Having reduced some aspect of the world to a form that can be
> processed according to rules, the output of the computation needs to
> be reintroduced into the world of meaningful, human action. To be put
> to work theoretically, information has to be reintegrated as
> meaningful and relevant evidence into a system of concepts,
> assumptions, and hypotheses. This involves translating the output
> into a familiar notational system and, in some cases, restoring more
> basic sensory modes of apprehension, as in the case of data
> visualization or the phenomenology of a thought experiment.

This he calls "expansion". Digital computation is involved in both the 
reduction (encoding, algorithmic analysis) and the expansion back into 
the human world (visualisation, mostly). In research, of course, the 
process is cyclical.

Viewed historically, e.g. in the tradition of triumphalist 
pronouncements e.g. from Lord Kelvin (1824-1907), Robert Millikan 
(1868-1953) and Ernest Rutherford (1871-1937), that "qualitative is 
just poor quantitative", Gooding counters that "there do appear to be 
modes of understanding that are inherently analogue rather than 
numerical or digital". He concludes:

> If humans are analogue devices, and it is humans that continue to do
> science, then we would expect the cyborg vision of the onward march
> of digitalization to need qualification. Instead of looking for
> cognitive capacities of the sort required by an algorithmic view of
> science as rule-based reasoning about an inherently digitizable
> world, we should investigate those cognitive capacities that enable
> practitioners from different cultures to exchange meanings and
> methods.



Willard McCarty (www.mccarty.org.uk/), Professor, Department of Digital
Humanities, King's College London, and Research Group in Digital
Humanities, University of Western Sydney

More information about the Humanist mailing list