Digit span is most often administered as part of an assessment
battery. The experiments presented here used the digit span test
from the Wechsler intelligence scales (e.g., Wechsler, 2008,
2009), though other standardized scales include similar digit span
tests (e.g., Elliot & Smith, 2011; Kaufman & Kaufman, 2004). These
tests generally assume that (forward) digit span measures
short-term verbal memory capacity as a basic cognitive mechanism.
This is somewhat troubling because studies that include digit
span – of which there are thousands – are in effect exaggerating
the role of STM capacity on the basis of digit span results, and playing
down the role of long-term instance-based associative learning
(or lack thereof). For example, links between dyslexia and poor
Experiment 2 used lists that were a mixture of digits and words
to show that isolated digits and isolated words were recalled
equally well but digit sequences were recalled more accurately
than word sequences. The British National Corpus was then used
in Experiment 3 to show that the superior recall of digit sequences
over word sequences arose because random sequences of digits
occurred more frequently than random sequences of words in natural
language. Furthermore, the model demonstrated that basic
associative mechanisms operating on the linguistic experience represented
by the corpus could account for the digit span advantage.
Experiment 4 then showed that pseudo-random digit lists containing
digit sequences that occur relatively frequently in the language
environment were recalled more accurately than pseudo-random
digit lists containing digit sequences that occur less frequently.
These results have specific implications for studies that have used
digit span; more generally though, they also have wider implications
concerning how long-term influences on STM are conceived.
Our findings suggest that the archetypal measure of verbal STM
capacity – that of digit span – is in part a function of the structure
of the natural linguistic environment. Indeed, if one begins to search for seemingly random digit sequences, their prevalence in the environment becomes startlingly
obvious. Using the UK as an example, the primary national
sport consistently lists results from matches as digit
sequences ; company phone numbers are often consistently repeated on
national television and radio advertisements as an aide memoire;
and bank account numbers and sort codes, used by almost every
adult, are series of random digits that are frequently used and
encountered. This provides an explanation for performance differences
that are seen across different stimulus sets.