www.elsevier.com/locate/paid

Personality and Individual Differences 39 (2005) 1005–1014

Working memory and general intelligence:The role of short-term storage

Roberto Colom a,*, Carmen Flores-Mendoza b,Ma Angeles Quiroga c, Jesus Privado a

a Facultad de Psicologıa, Universidad Autonoma de Madrid, 28049 Madrid, Spainb Av. Antonio Carlos, 6627, FAFICH-Depto. de Psicologia, Gab. 4042 UFMG, Cep: 31270-901 BH, Brazil

c Facultad de Psicologıa, Universidad Complutense de Madrid (Campus de Somosaguas), 28223, Madrid, Spain

Received 3 November 2004; accepted 9 March 2005

Available online 29 June 2005

Abstract

This study examined the relationships among short-term memory (STM), working memory (WM), and

general intelligence (g). Two independent samples of participants performed several verbal, quantitative,

and spatial STM and WM tasks, as well as a broad set of tests measuring psychometric intelligence. The

constructs are carefully sampled to include heterogeneous tasks and tests to control the effect of unwantedvariance. The results have several points of interest. First, there is a strong correlation between STM and

WM. Second, WM is a slightly better predictor of g than STM. Third, when the correlation between STM

and WM is statistically controlled, the unique predictive power of WM is small, which suggests that the

short-term storage component of the WM system largely drives the relationship between WM and g.

� 2005 Elsevier Ltd. All rights reserved.

Keywords: Memory span tasks; Short-term memory; Working memory; General intelligence

0191-8869/$ - see front matter � 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.paid.2005.03.020

* Corresponding author.

E-mail address: roberto.colom@uam.es (R. Colom).

1006 R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014

1. Introduction

The relationship between working memory (WM) and intelligence is broadly documented (Ack-erman, Beier, & Boyle, 2005). Several studies have considered the relationship between WM andan array of cognitive abilities: general intelligence or g (Ackerman, Beier, & Boyle, 2002; Colom,Abad, Rebollo, & Shih, in press; Colom, Rebollo, Palacios, Juan-Espinosa, & Kyllonen, 2004;Colom & Shih, 2004; Suß, Oberauer, Wittman, Wilhelm, & Schulze, 2002), reasoning ability(Kyllonen & Christal, 1990), fluid intelligence (Bayliss, Jarrold, Gunn, & Baddeley, 2003; Colom,Flores-Mendoza, & Rebollo, 2003; Conway, Cowan, Bunting, Therriault, & Minkoff, 2002;Engle, Kane, & Tuholski, 1999; Kane et al., 2004), spatial ability (Miyake, Friedman, Rettinger,Shah, & Hegarty, 2001), and reading comprehension (Daneman & Merikle, 1996). However, thecauses underlying the relationship between these constructs remain mysterious.WM tasks rely on the temporary maintenance of any given information while performing

some kind of concurrent processing. For instance, the reading span task requires the readingof several unrelated sentences and the temporary maintenance of the last word of each sentencefor later recall (Daneman & Carpenter, 1980). Presumably, those WM tasks are different fromtraditional short-term memory (STM) tasks, because the latter simply require the temporarymaintenance of several pieces of information without any explicit concurrent processingrequirement.Researchers have tried to answer the question of what make WM tasks predictors of perfor-

mance on cognitive ability measures. Daneman and Carpenter (1980) proposed the widely knownresource sharing model. Such a model postulates the existence of a trade-off between the storageand processing demands, WM tasks measure a functional capacity of resources that can be allo-cated between the storage and processing activities (Daneman & Hannon, 2001; Just & Carpenter,1992; Turner & Engle, 1989). However, Engle et al. (1999) claimed that WM capacity is deter-mined by STM capacity plus a so-called controlled attention ability. This ability is conceivedas a domain-general limited attentional capacity for performing controlled processing or sustain-ing focus on task-relevant information in the face of interfering or distracting stimuli. Interest-ingly, this model proposes that variance shared between WM and STM tasks should reflecttheir common short-term storage component, whereas the residual WM variance should reflectthe controlled attention or executive component of the WM system.Engle et al. (1999) have proposed that individual differences on measures of WM capacity pri-

marily reflect differences in controlled-attention capability. The factor underlying measures ofWM and higher level cognitive tasks (like performance on conventional intelligence tests) is theability to maintain a representation active in the face of interference from automatically activatedrepresentations competing for selection for action and in the face of distractions that would other-wise draw attention away from the currently needed representation.However, existing evidence indicates that the relation between WM and intelligence is not so

different from the relation between STM and intelligence. Mukunda and Hall�s (1992) meta-anal-ysis included articles published between 1976 and 1989. They distinguished WM and STM tasks,finding that the raw relationship between STM and intelligence was .26, whereas the raw relation-ship between WM and intelligence was .28. Ackerman et al. (2005) conducted a meta-analysisexamining the relationship between WM and intelligence, as well as between STM and intelli-gence. This study was based on studies published from 1872 to 2002. The meta-analytically

R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014 1007

derived correlation between ability and WM was .36, whereas the meta-analytically derived cor-relation between STM and ability was .28.Given that the correlations between STM and intelligence are in the same range as that of WM

and intelligence, it could be informative to consider the 3-way relationship among STM, WM, andintelligence, mainly because WM comprises short-term storage plus concurrent processing. Thevery few studies considering those relationships have taken a latent-variable approach. Engle,Tuholski, Laughlin, and Conway (1999), Conway et al. (2002), and Kane et al. (2004) found thatWM, but not STM, predict individual differences in intelligence. However, these studies have ta-ken three different operational approaches. When their datasets are analyzed against the same la-tent model, the results tell a different story: STM is a better predictor than WM when theircommon storage component is partialed out (Colom, Rebollo, Abad, & Shih, in press). Further,there are some instances in which STM and WM residuals predict individual differences in intel-ligence with the same power (see Colom, Rebollo, et al., in press). Finally, WM residuals are neverbetter predictors than STM.Nevertheless, most studies are limited in their assessment of the constructs of interest. Beier and

Ackerman (2004) noted that only two tests are usually taken as indicators of intelligence. Further-more, the indicators for WM and STM were also limited, because measures only from the verbalor from the spatial content domain are considered (Conway et al., 2002; Engle, Tuholski, et al.,1999; Miyake et al., 2001). It is widely accepted that when few measures are used to identify con-structs, their relations could be misrepresented because of specific test variance (Humphreys,1962). Thus, studies sampling the construct space through heterogeneous tasks to control forthe effect of unwanted variance are strongly required (the studies reported by Kane et al.(2004) as well as by Colom, Rebollo, et al. (in press), are noteworthy exceptions, but they havedelivered quite different conclusions).The purpose of the present study is to examine the simultaneous relationship among STM,

WM, and g. Ackerman et al.�s (2005) meta-analysis carefully considers the extreme importanceof the choice of tasks: ‘‘resolution of the question of how and how much WM and intelligenceare related ultimately requires additional research. In our opinion, the issue cannot be ultimatelysettled until studies are conducted that provide multiple tests of a wide range of ability factors(. . .), multiple tests of WM in each of the different content domains, (and) (. . .) WM tests thatdo not depend on time sharing performance’’. Furthermore, and especially germane for the pres-ent study, they consider crucial the principle of aggregation which indicates that multiple measuresof the same construct will yield a more stable and representative estimator than single behaviourmeasures. Aggregating across multiple measures will provide a more robust assessment of WM/STM/g relations, and might result in a more informative association.

2. Method

2.1. Participants

Two independent samples of Psychology undergraduates participated to fulfill a course require-ment. The first sample comprised 76 participants, whereas the second sample comprised 63 par-ticipants. Their mean age was 23.00 (SD = 5.5) and 20.9 (SD = 1.39) respectively.

1008 R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014

2.2. Measures

STM, WM, and g were measured through several tasks and tests. Verbal STM was measuredby the forward and backward letter span tasks, whereas quantitative STM was measured by theforward and backward digit span tasks. In order to make the tasks as simple as possible, we haveused single letters and single digits, instead of words or complex digits. Note that backward spantasks fit with other tasks of STM, because a simple transposition of order does not change itsnature (Cantor, Engle, & Hamilton, 1991; Rosen & Engle, 1997). Spatial STM was measuredby the corsi block and dot memory tasks, both requiring the maintenance of spatial information(location sequences and dot configurations, respectively), but not any explicit concurrent pro-cessing. The spatial STM tasks employed here modelled those administered by Miyake et al.(2001).Concerning WM tasks, it is important to note that not all were modelled after the reading

span task, because we wanted to define a WM composite score going beyond time sharing per-formance (Ackerman et al., 2005). Verbal WM was measured by the ABCD and the Alphabettasks. Both tasks were modelled after the CAM Battery (Kyllonen & Christal, 1990). The ABCDwas employed by Ackerman et al. (2002) as well as by Engle, Tuholski, et al. (1999) as a measureof verbal WM. Quantitative WM was measured by the Mental Counters and Computation Spantasks. The mental counters task was modelled after Larson and Saccuzzo (1989) and it was em-ployed by Mackintosh and Bennett (2003) as a measure of quantitative WM. The computationspan task was modelled after Ackerman et al. (2002). Finally, spatial WM was measured by thedot matrix and the letter rotation tasks. Both tasks were modelled after Miyake et al.�s (2001)study and they involve visuospatial storage (dot locations and spatial orientations, respectively)with a concurrent visuospatial processing (verification of spatial matrix equations or mentalrotation).A more detailed task description can be seen in Appendix A.Finally, the criteria for obtaining a fine-grained representation of the construct of general intel-

ligence were followed: number and variety of cognitive ability tests (Ackerman et al., 2002; Car-roll, 1993; Jensen & Weng, 1994). The test battery in the present study called for more than threeprimary factors (variety) which is clearly desirable to obtain a reliable composite score represent-ing g: ‘‘high-quality estimates of g are generated from the average across multiple tests of differingformats, contents, and processes’’ (Ackerman et al., 2005).The same STM and WM tasks were administered to both samples. However, the intelligence

tests were different. Five tests from the GATB battery were administered to the first sample:Series, Computation, Vocabulary, Arithmetic, and Spatial (Centro de Psicologia Aplicada,2001). The Culture Fair Intelligence Test was also administered (Cattell & Cattell, 2002).The second sample was assessed through three tests from the Differential Aptitude Battery(Verbal Reasoning, Spatial Relations, and Numerical Ability; Bennett, Seashore, & Wesman,1990), four tests from the Primary Mental Abilities Battery (vocabulary, mental rotation, com-putation, and inductive reasoning; Thurstone & Thurstone, 1939—computerized version byQuiroga, Privado, Beato, Colom, & Shih (2004)), the culture fair intelligence test (TEA,1997), the dominoes test (Anstey, 1955), and a quantitative reasoning test called ‘‘monedas’’(coins; Seisdedos, 1980).

R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014 1009

2.3. Procedure

The cognitive tasks were administered either individually or collectively in two sessions lastingone hour approx. each. The intelligence tests were administered collectively (lasting one hour forthe first sample and three hours for the second sample) following instructions given by the testmanuals.

3. Results

Before proceeding with the statistical analyses, all the measures were transformed to z-scores tocompute composites for the constructs of interest, namely, STM, WM, and g. For both samples,the STM composite score comprised six measures, as well as the WM composite score. The g com-posite for the first sample comprised six measures, whereas it comprised ten measures for the sec-ond sample. Therefore, presumably the obtained composite measures are reliable statisticalrepresentations of the analyzed constructs. The coefficient alpha across the scales in each compos-ite was: .80 and .80 (STM), .71 and .72 (WM), .73 and .87 (g) for the first and second samples,respectively.Firstly, the correlation between STM and WM was computed. The results revealed correlations

of .75 (confidence interval: .58 to .85) and .68 (confidence interval: .46 to .82) for the first and sec-ond samples, respectively. Thus, there is a strong relationship between the composite scoresreflecting short-term storage and storage + processing cognitive requirements.Secondly, regression analyses were applied to examine the straight relationship between STM

and g, as well as between WM and g in both samples. The results are shown in Table 1.WM is a better predictor of g than STM in both samples. However, the difference on the ob-

tained R values is rather small. This result is consistent with the view that the relationship betweenWM and g can be primarily explained by the short-term storage component of WM (Colom &Shih, 2004; Colom et al., in press; Colom, Rebollo, et al., in press).In addition to the previous analysis, two other smaller-sized composites were considered.

STM_A comprised forward letter and digit span, as well as the dot memory task, whereas STM_B

Table 1

R for STM and WM as predictors of g (confidence intervals are shown in parentheses at p < .01)

Sample A Sample B

STM WM STM WM

.52 .58 .50 .63

(.26 to .71) (.34 to .75) (.22 to .71) (.39 to .79)

STM_A WM_A STM_A WM_A

.47 .47 .41 .57

(.22 to .67) (.20 to .67) (.11 to .64) (.36 to .75)

STM_B WM_B STM_B WM_B

.49 .52 .41 .54

(.23 to .69) (.26 to .71) (.11 to .64) (.32 to .73)

All Rs are significant (p < .01).

1010 R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014

comprised backward letter and digit span, as well as the corsi block task. The later STM compos-ite includes two measures characterized by the transposition of order. WM_A comprised ABCD,computation span, and dot matrix, whereas WM_B comprised alphabet, mental counters, and let-ter rotation. The former WM composite includes two dual measures. These composites served aspredictors of g. The results are shown in Table 1. As expected, R values were smaller than thosecorresponding to composites computed after all the span measures. Nevertheless, the same pat-tern emerges: R values for STM and WM are not sharply different. Especially interesting is thefact that STM composites considering forward or backward span measures result in almost thesame R values, which reinforces the consideration of the backward versions as nice measuresof short-term storage. Further, the WM composite considering dual tasks predicts g with roughlythe same values as the other WM composite.Thirdly, given that WM is defined by storage + processing requirements, knowing the genuine

predictive power of WM (with its storage component partialed out) requires the computation ofWM residuals. Thus, regression analyses were applied to examine the relationship between STMand WM in order to estimate WM residuals (variance non-shared between WM and STM); WMwas predicted by STM. The resulting residuals are intended to reflect WM with its storage com-ponent partialed out.Hierarchical multiple regression was applied to examine the relationship among STM, WM

residuals, and g; STM was entered first. The results are shown in Table 2.The findings revealed that WM residuals add predictive power to that achieved by STM. R

values change from .52 to .59 in the first sample, whereas those values change from .49 to .67in the second sample. The change is statistically significant (p < .01) in both samples, which sug-gests that WM predicts g beyond its temporary storage component Nevertheless, the contribu-tion of WM residuals is not especially noteworthy: .07 and .18 for the first and second samplesrespectively.Finally, the smaller-sized composites considered in Table 1 are taken as predictors of g on a

hierarchical multiple regression analysis. The results depicted in Table 2 are comparable to thoseobtained from the composites comprising all the span measures: the contribution of WM residualsis low (.08 and .09 for the first sample, and .22 and .15 for the second sample). Therefore, WMresiduals must be considered germane to predict g, but most of the predictive power that canbe attributed to WM derives from its short-term storage component.

Table 2

Multiple R for STM and WM residuals as predictors of g (confidence intervals are shown in parentheses at p < .01)

Sample A Sample B

STM WM(r) STM WM(r)

.52 .07 .49 .18

(.26 to .71) (�.23 to .36) (.22 to .71) (�.14 to .47)STM_A WM_A(r) STM_A WM_A(r)

.47 .08 .41 .22

(.20 to .67) (�.22 to .37) (.11 to .64) (�.11 to .50)STM_B WM_B(r) STM_B WM_B(r)

.49 .09 .40 .15

(.23 to .69) (�.21 to .38) (.09 to .64) (�.18 to .45)

R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014 1011

4. General discussion

The present study considered a broad and (presumably) representative sample of memory spanmeasures. Several measures are thought to represent STM, whereas the remaining are thought torepresent WM. From a theoretical standpoint, WM is supposed to comprise storage plus process-ing requirements, whereas STM comprises mainly storage. Several cognitive ability tests were alsoadministered to obtain a fine-grained representation of g.The results have several points of interest. First, there is a strong relationship between STM and

WM. The main implication is that cognitive tasks characterized by short-term storage require-ments are not sharply distinguishable from cognitive tasks requiring storage plus processingrequirements. This finding is consistent with previous studies that have taken a latent-variable ap-proach. Engle et al. (1999) reported a correlation of .68 between STM and WM. Conway et al.(2002) reported a correlation of .82. Colom et al. (in press) found a correlation of .89. Colomet al.�s (in press) re-analysis of the Kane et al.�s (2004) dataset revealed a correlation of .99 be-tween those latent constructs. Colom, Abad, Shih, and Aguilera (submitted for publication) founda correlation of .87. Finally, Miyake et al. (2001) reported a correlation of .86 between STM andWM. Those results support the view that STM and WM do not reflect clearly distinguishable cog-nitive limitations.Second, the composite score representing STM predicted the composite score representing g. In

fact, R values connecting STM and g were not very different from R values connecting WM and g,which suggests that the storage + processing combination is not crucial to account for the corre-lation between WM and g.Third, the results of the hierarchical regression analyses revealed that the WM composite resid-

ual score (with its temporary storage component partialed out) adds predictive power to the ob-served relationship between STM and g which suggests that the relationship between WM and gcannot be fully explained by the temporary storage component of the WM system. The finding isconsistent with the study reported by Colom et al. (in press) who found that STM andWM (withits storage component partialed out) predict g. Nevertheless, the contribution of the non-storagecomponents of WM cannot be considered especially noteworthy: WM residuals add little to theprediction achieved by short-term storage only.There are some researchers arguing that executive functioning underlies the relationship be-

tween WM and intelligence (Engle & Kane, 2004). It seems reasonable to state that WM residualsrepresent executive functioning. However, Colom et al. (submitted for publication) found thatshort-term storage and processing efficiency predict the relationship between WM and g. Theirlatent-variable analyses showed that when storage and processing components are extracted fromWM, the correlation between the resulting WM residual and g is not significantly different fromzero. Furthermore, STM was a much better predictor of WM than processing efficiency.Taken together, the results are consistent with the view that the short-term storage component of

the WM system is a crucial factor underlying the relationship between WM and g (Colom & Shih,2004). Although temporary storage does not exhaust the predictive power of WM, it can be statedthat this component is much more important than some theoretical models assume (Engle & Kane,2004). The present study adds evidence supporting the view that WM components distinguishablefrom short-term storage can be germane to account for the relationship between WM and g, buttheir contribution to the prediction of g is much less compelling than previously assumed.

1012 R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014

Acknowledgement

The research referred to in this article was supported by a grant funded by the Spanish ‘‘Min-isterio de Ciencia y Tecnologıa’’ (Grant No. BSO2002-01455).

Appendix A

Forward Letter Span (FLSPAN), Backward Letter Span (BLSPAN), Forward Digit Span(FDSPAN), and Backward Digit Span (BDSPAN). Single letters or digits (from 1 to 9) were pre-sented on the computer screen at the rate of one letter or digit per second. Unlimited time wasallowed to type in direct or reverse order the letters or digits presented. The set size ranged fromthree to nine (7 levels · 3 trials each = 21 trials total). The score was the number of accuratelyreproduced trials.Corsi block. Nine boxes are shown on the computer screen and one box at a time turned orange

for 650 ms each. Immediately after a sequence of taps, their order must be repeated. The se-quences increased from 3 to 9 taps (7 levels · 3 trials each = 21 trials total). The score was thenumber of trials reproduced appropriately.Dot memory. One 5 · 5 grid was displayed for 750 ms at the computer screen. Each grid had

between two and seven spaces comprising solid dots. After the grid presentation, the locationsthat contained dots must be recalled. The trials increased progressively from two to seven dots(6 levels · 3 trials each = 18 trials total). The score was the number of trials correctly reproduced.ABCD. Two categories and five words in each category were used. Three study frames were dis-

played on the computer screen for three seconds each. The first frame indicated the order of twomembers from one category (Cedar before Oak), the second frame indicated the order of twomembers from the other category (Garlic not Before Salt) and the third frame indicated the orderof the categories (Trees not Before Food). After the third study screen, an eight choice answerscreen was presented in order to select the correct order of the words obtained from the informa-tion given by the displayed sentences. There were 14 trials. The score was the number of correcttrials.Alphabet. This task required the application of successor and predecessor operations to a string

with a given number of letters. The trials increased the number of letters from three to seven (5levels · 4 trials each = 20 trials total). For two trials within a given block 1 or 2 positions mustbe added, while for the other two trials 1 or 2 positions must be subtracted. The score was thenumber of correct trials.Mental counters. Three boxes representing counters appeared on the computer screen. A yellow

star appeared above or below one counter for 500 ms. If the star appeared above the box, onemust be added (+1) to that counter, but if the star appeared below the box, one must be sub-tracted (�1) to that counter. The task requires keeping a running track of the value of the threecounters. The cumulative total of all three counters must be reported. 10 trials comprised fivecounter changes, whereas 10 trials comprised seven counter changes. The score was the numberof correct trials.Computation span. This task included a verification task and a recall task. The accuracy of a

math equation must be verified and the displayed solution, irrespective of its accuracy, must be

R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014 1013

remembered. The solutions from the equations must be remembered in their correct serial order.The trials ranged from three to seven equation/solutions (5 levels · 3 trials each = 15 trials total).The participant�s score was the number of hits in the verification and remembering tasks.Dot matrix. A matrix equation must be verified and then a dot location displayed in a 5 · 5 grid

must be temporarily retained. The matrix equation requires adding or subtracting simple linedrawings. Once the response is delivered, the computer displayed the grid for 1.5 s. After a givensequence of equation-grid pairs, the grid spaces that contained dots must be recalled. The trialsincreased in size from two to five equations and dots (4 levels · 3 trials = 12 trials total). The scorewas the number of hits in the verification and remembering tasks.Letter rotation. Several uppercase letters are presented sequentially normal or mirror imaged.

They can be rotated in one of seven orientations (multiples of 45�). There is a verification task(is the letter normal or mirror imaged?) and a recall task (the orientation of the displayed let-ters—where was the top of each letter pointing?). After each set a grid was depicted to markthe places corresponding to the positions of the tops of the presented letters in their correct serialorder. The trials increased in size from two to five letters (4 levels · 3 trials, 12 trials total). Thescore was the number of hits in the verification and remembering tasks.

References

Ackerman, P. L., Beier, M. E., & Boyle, M. O. (2002). Individual differences in working memory within a nomological

network of cognitive and perceptual speed abilities. Journal of Experimental Psychology: General, 131(4), 567–589.

Ackerman, P. L., Beier, M. E., & Boyle, M. O. (2005). Working memory and intelligence: the same or different

constructs? Psychological Bulletin, 131(1), 30–60.

Anstey, E. (1955). D-48. Tests de domino [Dominoes test, 1996]. Madrid: TEA.

Bayliss, D. M., Jarrold, C., Gunn, D., & Baddeley, A. (2003). The complexities of complex span: explaining individual

differences in working memory in children and adults. Journal of Experimental Psychology: General, 132(1), 71–92.

Beier, M. E., & Ackerman, P. L. (2004). A reappraisal of the relationship between span memory and intelligence via

‘‘best evidence synthesis’’. Intelligence, 32(6), 607–619.

Bennett, G. K., Seashore, H. G., & Wesman, A. G. (1990). Differential aptitude test (5th ed., 2000). Madrid: TEA, S.A.

Cantor, J., Engle, R. W., & Hamilton, G. (1991). Short-term memory, working memory, and verbal abilities: how do

they relate? Intelligence, 15, 229–246.

Carroll, J. B. (1993). Human cognitive abilities. Cambridge: Cambridge University Press.

Cattell, R. B., & Cattell, A. K. S. (2002). Teste equicultural de intelig~encia. adaptado pelo centro de psicologia aplicada[Culture fair intelligence test]. Rio de Janeiro: CEPA.

Centro de Psicologia Aplicada (2001). Bateria de Testes de Aptidao Geral -BTAG II [GATB Battery]. Adapted by

Gustavo Lisboa Braga. Rio de Janeiro: CEPA.

Colom, R., Abad, F. J., Rebollo, I., & Shih, P. C. (in press). Memory span and general intelligence: a latent-variable

approach. Intelligence.

Colom, R., Abad, F. J., Shih, P. C., & Aguilera, A. (submitted for publication). Short-term memory and processing

speed predict the relationship between working memory and g.

Colom, R., Flores-Mendoza, C., & Rebollo, I. (2003). Working memory and intelligence. Personality and Individual

Differences, 34, 33–39.

Colom, R., Rebollo, I., Palacios, A., Juan-Espinosa, M., & Kyllonen, P. (2004). Working memory is (almost) perfectly

predicted by g. Intelligence, 32, 277–296.

Colom, R., Rebollo, I., Abad, F. J., & Shih, P. C. (in press). Complex span tasks, simple span tasks, and cognitive

abilities: a re-analysis of key studies. Memory and Cognition.

Colom, R., & Shih, P. (2004). Is working memory fractionated onto different components of intelligence? Intelligence,

32, 431–444.

1014 R. Colom et al. / Personality and Individual Differences 39 (2005) 1005–1014

Conway, A. R. A., Cowan, N., Bunting, M. F., Therriault, D. J., & Minkoff, S. R. B. (2002). A latent variable analysis

of working memory capacity, short-term memory capacity, processing speed, and general fluid intelligence.

Intelligence, 30, 163–183.

Daneman, M., & Carpenter, P. A. (1980). Individual differences in working memory and reading. Journal of Verbal

Learning and Verbal Behaviour, 19, 450–466.

Daneman, M., & Hannon, B. (2001). Using working memory theory to investigate the construct validity of multiple-

choice reading comprehension tests such as the SAT. Journal of Experimental Psychology: General, 130, 208–223.

Daneman, M., & Merikle, P. M. (1996). Working memory and language comprehension: a meta-analysis. Psychonomic

Bulletin & Review, 3, 422–433.

Engle, R. W., & Kane, M. J. (2004). Executive attention, working memory capacity, and a two-factor theory of

cognitive control. In B. Ross (Ed.). The psychology of learning and motivation (Vol. 44, pp. 145–199). NY: Elsevier.

Engle, R. W., Kane, M. J., & Tuholski, S. W. (1999). Individual differences in working memory capacity and what they

tell us about controlled attention, general fluid intelligence, and functions of the prefrontal cortex. In A. Miyake &

P. Shah (Eds.), Models of working memory (pp. 102–134). New York: Cambridge University Press.

Engle, R. W., Tuholski, S. W., Laughlin, J. E., & Conway, A. R. A. (1999). Working memory, short-term memory, and

general fluid intelligence: a latent-variable approach. Journal of Experimental Psychology: General, 128(3), 309–331.

Humphreys, L. G. (1962). The organization of human abilities. American Psychologist, 17(7), 475–483.

Jensen, A., & Weng, L. (1994). What is a good g? Intelligence, 18, 231–258.

Just, M. A., & Carpenter, P. A. (1992). A capacity theory of comprehension: individual differences in working memory.

Psychological Review, 99, 122–149.

Kane, M. J., Hambrick, D. Z., Tuholski, S. W., Wilhelm, O., Payne, T. W., & Engle, R. W. (2004). The generality of

working memory capacity: a latent-variable approach to verbal and visuo-spatial memory span and reasoning.

Journal of Experimental Psychology: General, 133, 189–217.

Kyllonen, P. C., & Christal, R. (1990). Reasoning ability is (little more than) working memory capacity?! Intelligence,

14, 389–433.

Larson, G. E., & Saccuzzo, D. P. (1989). Cognitive correlates of general intelligence: toward a process theory of g.

Intelligence, 13, 5–31.

Mackintosh, N. J., & Bennett, E. S. (2003). The fractionation of working memory maps onto different components of

intelligence. Intelligence, 31, 519–531.

Miyake, A., Friedman, N. P., Rettinger, D. A, Shah, P., & Hegarty, M. (2001). How are visuospanal working memory,

executive functioning, and spatial abilities related? A latent-variable analysis. Journal of Experimental Psychology:

General, 130(4), 621–640.

Mukunda, K. V., & Hall, V. C (1992). Does performance on memory for order correlate with performance on

standardized measures of ability? A meta-analysis. Intelligence, 16, 81–97.

Quiroga, Mª. A., Privado, J., Beato, Mª. S., Colom, R., & Shih, P. C. (2004). Test de aptitudes mentales primarias,

Version informatica [Primary mental abilities battery, computerized version]. Madrid: TEA.

Rosen, V. M., & Engle, R. W. (1997). Forward and backward serial recall. Intelligence, 25, 37–47.

Seisdedos, N. (1980). Monedas [Coins]. Madrid: TEA.

Suß, H., Oberauer, K., Wittman, W., Wilhelm, O., & Schulze, R. (2002). Working memory capacity explains reasoning

ability–and a little bit more. Intelligence, 30, 261–288.

TEA (1997). Test de factor g de Cattell [The cattell�s culture-fair intelligence test]. Madrid: TEA.Thurstone, L. L., & Thurstone, Th. G. (1939). Primary metal abilities [1996]. Madrid: TEA.

Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language,

28, 127–154.