The Dictionary of Human Geography

Comp. by: VPugazhenthi Stage : Revises1 ChapterID: 9781405132879_4_N Date:31/
3/09 Time:15:13:07 Filepath:H:/00_Blackwell/00_3B2/Gregory-9781405132879/appln/
3B2/revises/9781405132879_4_N.3d

their different main constituents/actors, types

of relation foregrounded and methods of

analysis:

(1) Infrastructural technically based net-

works, such as electrical, road, rail, sew-

erage and telecommunications systems,

can be described according to their

density, connectedness and orientation

(Haggett, 1969: seeinfrastructure).

(2) social networks, such as kinship,

friendship and communities, have histor-

ically been analysed both quantitatively

by social network analysis and more

qualitatively using tools derived from

social anthropology (Strathern, 1996).

(3) Network-based models of organization

have tended to merge the distinctive fea-

tures of the previous two approaches, as

the nature of collectives from the infor-

mal and local to formal and global are

increasingly seen as exhibiting this kind

of form (Castells, 1996b; Barry, 2001).

(4) Finally, actor-networks (Latour, 1993)

are the distributed forms of agency that

emerge from articulation of humans and

non-humans as seen by practitioners of

the conceptual approach that originated

in science and technology studies

(STS) and is known asactor-network

theory. nb

neural networks Methods of finding solu-

tions to a range of technical problems using

computeralgorithmsbased on models of the

human brain. The networks are ‘trained’ to

find solutions to new problems by being

given exemplars on which to base their de-

cisions, such as which category to assign an

individual observation to when the category

boundaries are fuzzy (cf.fuzzy sets). The

approach is particularly valuable in various

forms of pattern recognition and classification

and is widely used inremote sensingstudies

for classifying segments of the Earth’s surface

according to theirland cover(Foody, 1996).

The spectral signatures of different land-cover

types vary, and many small areas for which

data are obtained contain a mixture of types.

The goal is to allocate each observation unit

(pixel) to a relatively homogeneous category.

‘Training’ sites are defined, using synthetic

pixels with homogeneous land cover, and the

neural network algorithm, through an iterative

procedure, allocates all of the observed pixels

to the type it most closely resembles. It is then

possible to estimate the proportion of the land

surface under different types of cover.

A similar approach has been suggested for

classificationof socio-economic data. Very

few small areas – such ascensus tracts– are

homogeneous in their population character-

istics so clear boundaries between types of

area cannot be defineda priorion other than

pragmatic grounds (e.g. use of quartiles); they

are the same as pixels with mixed land

uses. Using neural network approaches,

ideal typesare defined (with different levels

of homogeneity and mixtures, for example)

and the individual tracts are allocated to

those they most resemble (Mitchell, Martin

and Foody, 1998). rj

Suggested reading

Openshaw and Openshaw (1997).

New Economic Geography An approach as-

sociated primarily with a group of American

neo-classical economists (seeneo-classical

economics) who from the early 1990s sought

to apply theoretical rigour, analytical methods

and econometric (statistical) techniques to a

space-economy. Such interest was surprising,

given economists’ historical attachment to

an a-spatialeconomy, to ‘a wonderland of

no dimensions’ (Isard, 1956, p. 25). For Paul

Krugman (1995a, p. 33), however, the econo-

mist most central to the movement, the New

Economic Geography was ‘a vision on the road

to Damascus’: ‘I suddenly realised that I had

spent my whole professional life .. .thinking

and writing about economic geography, with-

out being aware of it’ (Krugman, 1991, p. 1).

Krugman’s epiphany was that the analytical

framework he previously deployed to under-

stand internationaltradewas perfect for com-

prehending economic geography: ‘Economic

geography, like .. .trade theory, is largely

about increasing returns and multiple equilib-

ria. The technical tricks needed to make

models tractable are often the same’

(Krugman, 1995b, p. 41). That term ‘model’

is critical. While conventional economic geog-

raphy was ‘a field full of empirical insights,

good stories and obvious practical import-

ance’, it was ‘neglected’ – that is, neglected

by economists – ‘because nobody had seen a

good way to formalize it’ (Krugman, 1995b,

p. 41). Krugman’s project was to take works of

economic geographers and to make them (as

he saw it) intellectually viable by expressing

them in the formal vocabulary of economic

models. As he writes, ‘we will integrate

spatial issues into economics through clever

models .. .that make sense of the insights of

the geographers in a way that meets the formal

Gregory / The Dictionary of Human Geography 9781405132879_4_N Final Proof page 499 31.3.2009 3:13pm Compositor Name: ARaju

NEW ECONOMIC GEOGRAPHY