Implications of Infant Cognition for Executive Functions at Age 11

Abstract

Recent work suggests that executive functions, the cornerstone of higher-level cognitive operations, are driven by basic information processing abilities. Using structural equation modeling, with latent variables, the present study provides the first evidence that this driving force begins in infancy, such that abilities in infancy predict executive functions at age 11. Information processing abilities in three domains (attention, processing speed, and memory) were assessed when participants were infants (7 and 12 months) and toddlers (24 and 36 months) and were used to predict three executive functions (working memory, inhibition, and shifting) when participants were 11 years old. A model relating infant abilities to age-11 executive functions fit well, and accounted for 9% to 19% of the variance in the executive functions. Paths from both Speed and Memory in infancy to age-11 Working Memory were significant, as was the path from Speed in infancy to age-11 Shifting. A model using abilities in toddlerhood as predictors fit similarly. These findings implicate early basic cognitive abilities in the development of executive functions.

Keywords

infant cognition information processing executive function adolescent cognition

Executive functions (EFs) are higher-level cognitive abilities needed for social and behavioral competence and academic achievement (Jurado & Rosselli, 2007). They encompass a wide variety of skills that undergird purposeful, goal-directed activity, including self-regulation, planning, problem solving, and flexible thinking. Among the most widely studied EFs are the three interrelated but distinct abilities identified by Miyake, Friedman, Emerson, Witzki, and Howerter (2000): updating (or working memory), which refers to holding and manipulating information in short-term memory; inhibition, the deliberate suppression of dominant, automatic, or prepotent responses when necessary; and shifting, the flexible switching of attention between tasks, strategies, or mental sets. These three EFs are often considered foundational, because of their involvement in more complex EF tasks.

The Role of Basic Information Processing Abilities in Executive Functions

Recent studies with school-age children and adults suggest that basic information processing abilities—for example, processing speed, attention, and memory—underlie the pronounced individual differences often seen in EFs, as well as the prominent age-related changes in EFs.

Current evidence indicates that processing speed plays a key role (Kail, 2007; Salthouse, 1996). In adults, age-related declines in speed have been found to account for age-related declines in working memory (Salthouse, 2005) and individual differences in shifting (Salthouse, 2011). In children, developmental improvements in processing speed account for age-related improvements in working memory (Kail, 2007). In a recent study from our lab, using measures obtained when participants were 11 years old, we found that processing speed mediated differences in working memory, shifting, and inhibition between participants who had been born preterm and those who were born full-term (Rose, Feldman, & Jankowski, 2011). Moreover, when processing speed was partialed from the three EFs, their interrelation was substantially reduced, a result further substantiating the pivotal role played by speed.

Attention, and particularly selective attention, is also thought to influence EFs. Selective attention has been shown to constrain both working memory (Cowan, 2011) and inhibition (Kane & Engle, 2003). Evidence for the role of attention in working memory comes largely from studies showing that adults with low working memory capacity have difficulty ignoring distractors. In children, age-related increases in attention efficiency, indexed by the ability to filter out irrelevant information, are associated with age-related improvements in working memory (Cowan, 2011).

Most studies relating basic abilities to EF have focused on but a single basic ability, and often on only a single EF. However, in an omnibus study, with two very large samples of adults, Salthouse (2005) examined the relation of five basic information processing abilities to multiple EFs. The findings showed that three of these abilities—perceptual speed, memory, and reasoning—were implicated in a wide range of EFs, including working memory, inhibition, and measures of more complex functions (e.g., planning, problem solving, and verbal fluency).

Basic Abilities in Infants and Toddlers

Studies dealing with the emergence and development of basic information processing abilities have figured prominently in the recent literature on infants. A number of abilities similar to those found in older children and adults have been putatively identified in infancy; these include processing speed, attention, memory, and aspects of abstraction and representation involved in reasoning (for an overview, see Rose & Tamis-Lemonda, 1999).

There are two reasons to entertain the notion that these early-appearing abilities might have long-term implications for complex cognitive functions. First, although cognitive abilities are assessed very differently in infants and adults, mounting evidence indicates that there are significant continuities from infancy to later years. In one study involving preterm infants, neonatal attention (indexed by look duration) was found to relate to selective attention at age 12 years (Sigman, Cohen, Beckwith, Asarnow, & Parmelee, 1991). Another study measured ocular reaction time (RT) to targets in a visual expectation paradigm and found that this index of processing speed at age 3.5 months related to ocular RT at age 4.5 years (Dougherty & Haith, 1997). Furthermore, several studies found that infant recognition memory (indexed by novelty scores on the paired-comparison task) was related to standard assessments of memory through at least 11 years (Rose & Feldman, 1995), and another found that recall memory at age 20 months (indexed by scores on the elicited-imitation task) was related to standard assessments of recall at age 6 years (P. J. Bauer, personnel communication, May 23, 2011). Finally, in a comprehensive longitudinal study assessing information processing at five different ages, significant continuities from infancy and toddlerhood to age 11 years were found in several domains, including attention, processing speed, and memory (Rose, Feldman, Jankowski, & Van Rossem, 2012). The existence of such continuities lends credence to the idea that the discrete abilities found in infancy are early instantiations of their more mature counterparts.

Second, there is evidence from developmental neurobiology pointing to similarities in the neural underpinning of infant and adult abilities, particularly those involved in recognition and recall memory. For one thing, despite the methodological differences between tasks of infant and adult memory, performance relies on the same brain structures of the medial temporal lobe (Squire, Wixted, & Clark, 2007). Additionally, adults with amnesia due to medial temporal lobe damage perform poorly on tasks used to assess infants’ recognition and recall (McDonough, Mandler, McKee, & Squire, 1995; McKee & Squire, 1993), as do monkeys with lesions to these same areas (Bachevalier, Brickson, & Hagger, 1993). These findings are buttressed by results from a recent latent-variable structural equation modeling (SEM) study with infants and toddlers (Rose, Feldman, Jankowski, & Van Rossem, 2011) showing that memory is structured the same way at these young ages as it is in adults (Quamme, Yonelinas, Widaman, Kroll, & Suave, 2004); specifically, recognition is supported by two processes—familiarity and recollection—whereas recall is supported by only one—recollection. Moreover, recollection, but not familiarity, is affected by risk for hippocampal damage in infants, as in adults.

Because measures of information processing in infants have been found to relate to complex cognitive abilities, such as IQ (Fagan, Holland, & Wheeler, 2007; Rose & Feldman, 1995), the possibility arises that they may also relate to EFs. However, the relation between IQ and EFs is neither clear nor straightforward. Although deficits in EFs are frequently associated with frontal lobe damage, individuals with such damage often have normal IQs (Damasio, 1994). Moreover, Friedman et al. (2006) found that not all EFs were related to IQ even in normal populations: Working memory was, but shifting and inhibition were not. In short, the relation between information processing in infancy and later EFs is still an open question.

The Present Study

The question of central interest in the present study was whether basic information processing abilities in infancy (7 and 12 months) and toddlerhood (24 and 36 months) relate to later EFs (age 11 years). The basic abilities were drawn from three domains—processing speed, attention, and memory—and the same measures, in the same formats, were used at the infant and toddler assessments. The age-11 battery included assessments of the three EFs identified by Miyake et al. (2000)—working memory, inhibition, and shifting—and employed tasks typically used with older children and adults. The study addressed two major questions: Do any of these basic processes, when assessed in infancy and toddlerhood, relate longitudinally to EFs at age 11 years? If so, are any of the infant or toddler measures differentially related to age-11 EFs? Confirmatory factor analysis (CFA) and latent-variable path analysis were used to examine these issues. This study is unique in using core cognitive abilities measured during infancy to predict later EFs. Findings from this study could substantially extend current knowledge regarding the long-term implications of infant cognition.

Method

Participants

The sample was drawn from a prospective longitudinal study of cognitive development that included both preterm and full-term infants. The mean age at EF testing was 11.18 years (SD = 0.44) for the preterm group and 11.14 years (SD = 0.35) for the full-term group. (As discussed in the Data Analysis section, because relations among measures, including those between early and later measures, did not differ between the two groups, their data were combined for the analyses reported here.)

Original sample

The original sample included 203 children (59 preterm infants and 144 full-term control infants), born between February 1995 and July 1997. Preterm infants admitted to the neonatal intensive-care units of two hospitals affiliated with Albert Einstein College of Medicine were recruited. Criteria for inclusion in the study were singleton birth, birth weight below 1,750 g, gestational age less than 37 weeks, and the absence of any obvious congenital, physical, or neurological abnormalities. Term infants were recruited from the same hospitals; criteria for study intake were birth weight above 2,500 g, gestational age of 38 to 42 weeks, 5-min Apgar score of 9 or 10, and uneventful pre- and perinatal circumstances. Before the follow-up at 11 years, the children had been seen at ages 5, 7, 12, 24, and 36 months.

Attrition

Of the original 203 children, 134 returned when they were 11 years old; the follow-up rate was 74.6% for the preterm group (n = 44) and 62.5% for the full-term group (n = 90). This attrition appears reasonable in view of the 8-year period since the last visit. Reasons for loss to follow-up included families moving out of state (n = 24), inability to locate families that had moved (n = 35), and refusal to participate (n = 7). Data for children who developed serious neurological conditions since their last follow-up, at age 3, were excluded (n = 3, all from the full-term group).

Background and medical characteristics

Background factors were similar for the preterm and full-term groups; overall, 50.7% of the participants were male, 36.0% were first born, and 86.2% were either Black or Hispanic. Maternal education averaged 13.6 years (SD = 2.1), and SES as assessed with the Hollingshead Four-Factor Index (Hollingshead, 1975) averaged 36.7 (SD = 13.0). Background factors for children returning at age 11 were nearly identical to those of the original sample (for more details, see Rose, Feldman, & Jankowski, 2001). Medical risk factors of members of the preterm group who returned at age 11 were also similar to those of the original cohort.

Procedure: information processing tasks in the infant and toddler periods

The information processing measures for infants (7 and 12 months) and toddlers (24 and 36 months) assessed performance in three domains: memory, processing speed, and attention (see Table 1). Memory was assessed with two immediate-recognition tasks using the paired-comparison procedure, a delayed-recognition task measuring ability to distinguish a novel stimulus from one presented during a habituation period, and a recall task involving elicited imitation. Processing speed was assessed with a psychomotor task that measured ocular RT to briefly presented stimuli and an encoding-speed task that measured the rapidity with which information about targets was encoded. Attention was assessed with measures of look duration and shift rate culled from a number of tasks. (Shorter look durations and faster shift rates are associated with better attention.) Tasks were made age appropriate as the infants grew older by shortening presentation and test times, increasing the stringency of learning criteria, or increasing stimulus complexity (for more details, see Rose, Feldman, & Jankowski, 2005a, 2005b, 2009; Rose, Feldman, Jankowski, & Van Rossem, 2008).

Table 1.

Summary of the Information Processing Tasks Administered to Infants and Toddlers

Ability and task	Description	Dependent variable	Reliability
Memory
Immediate recognition (two batteries)	On each two-part trial, familiarization to one target was followed by a test presentation of the familiar target paired with a novel one.	Novelty score (percentage of time looking to the novel test target)	r = .79–.80 (correlation between performance on the two batteries)
Delayed recognition	Following habituation to a series of objects and after delays of 1, 3, or 5 min, familiar objects were presented paired with novel ones.	Novelty score (percentage of time looking to the novel test target)	α = .30–.39
Recall	Following modeling of a series of four sequences of events (one at a time) and a 15-min delay, the child was given props for each event, in turn, and encouraged to reproduce it.	Percentage of target actions reproduced correctly	r = .52–.77 (split-half reliability: unequal-length Spearman-Brown formula)
Processing speed
Psychomotor speed (reaction time)	Computer-presented stimuli appeared briefly, some to the left (L) and others to the right (R) of the midline. The location of these targets was random on 10 baseline trials and then followed a predictable (R-R-L) sequence for 60 trials.	Mean reaction time (on trials in which the location of the target changed; reaction times < 150 ms excluded)	r = .62–.88 (odd-even split-half reliability)
Encoding speed	Paired targets were presented; one target changed from trial to trial, and the other remained constant.	Number of trials to criterion (consistent preference, as indexed by looking time, for the novel target)	r = .40–.50 (test-retest reliability)
Attention
Look duration	The average duration of individual looks to targets was culled from a number of tasks (e.g., immediate recognition, encoding speed)	Composite score formed by standardizing and then averaging looking-time scores (shorter looks indicate better attention)	α = .68–.81
Shift rate	The number of shifts of gaze between paired targets was culled from several tasks (e.g., immediate recognition, encoding speed)	Composite score formed by standardizing and then averaging shift rates (higher rates indicate better attention)	α = .66–.71

Note: All tasks were administered when participants were 7, 12, 24, and 36 months old, except that the recall task was not administered at 7 months. All alpha coefficients are standardized.

The early measures of memory, processing speed, and attention have previously been found to form separate constructs (Rose, Feldman, & Jankowski, 2004, 2005b)¹ and to relate to more mature instantiations of the same constructs in preadolescence (Rose et al., 2012).

Procedure: measures of executive function at age 11

The EF measures (see Table 2) assessed three components of executive functioning: working memory, inhibition, and shifting. Two types of working memory were examined: simple (storage capacity) and complex (storage capacity plus concurrent mental operations).² Although these two aspects of working memory sometimes load on separate factors (Gathercole, Pickering, Ambridge, & Wearing, 2004), they are not necessarily distinct in early childhood and adolescence, when the simple tasks may draw on some of the same executive resources as more complex ones (Unsworth & Engle, 2006).

Table 2.

Summary of the Executive Function Tasks Administered at Age 11 Years

Ability and task	Description	Dependent variable	Reliability^a
Working memory: storage
Spatial span (CANTAB)	Nine boxes appeared in random locations on the screen, and an increasing number lit up on each successive trial.	Longest sequence correctly reproduced	r = .64
Pattern span	Increasingly complex matrices of white and black squares were presented.	Largest matrix reproduced	r = .75
Change detection	Arrays of 2, 4, 6, or 8 colored squares appeared briefly (250 ms); on successive presentations, either the array remained the same or the color of one constituent square changed.	Number of correct same/ different judgments and latency to respond	α = .78
Working memory: storage and processing
Spatial working memory (CANTAB)	Four, 6, or 8 boxes were arrayed on a computer screen, with a token hidden under each. The child had to keep the location of previously searched boxes in mind to retrieve all the tokens in a minimum number of tries.	Number of errors (searching the same box more than once)	r = .68
Counting span	The number of red circles in an array was counted (ignoring distractors), and the sum was restated. Arrays were presented in sets, and the number of arrays per set increased from 2 to 5. After each set, the series of sums was to be recalled in order.	Proportion of items recalled in the correct serial order per set size (a partial-credit unit-weighted score)	α = .58
Listening span	After listening to a sentence, the child judged its veracity (true/false) and repeated its final word. Sentences were presented in sets, and the number of sentences per set increased from 2 to 6. After each set, the series of final words was to be recalled in order.	Proportion of items recalled in the correct serial order per set size (a partial-credit unit-weighted score)	α = .70
Inhibition
Go/no-go	Eight Looney Tunes cartoon characters appeared on the screen. The child pressed a key to indicate the appearance of any of seven of the characters (go trials) and withheld a response to Bugs Bunny, the remaining character (no-go trials).	Number of correct no-go trials	α = .81
Rapid visual information processing (CANTAB)	Single digits appeared in a stream in quick succession (100/min) on a monitor. The child was to press a button only when the target sequence (3-5-7) appeared.	Number of false alarms (errors of commission)	r = .74
Shifting
Trail making	Twenty-five small circles, each containing a target, were to be connected in sequence as quickly as possible: In Part A (baseline), targets were all single digits; in Part B, targets alternated between numbers and letters (1, A, 2, B, 3, C, etc.).	Difference in time to complete Parts A and B	r = .80–.90
Intradimensional- extradimensional shift (CANTAB)	Paired shapes were presented, and the child learned through trial and error which shape in each pair to choose. Following initial learning, there were several series of rule changes governing which shape was correct. Some changes were intradimensional (a new exemplar of a previously reinforced dimension was rewarded), and others were extradimensional (a new dimension was rewarded).	Percentage of errors on reversal blocks	r = .70

Test-retest reliabilities for the Cambridge Neuropsychological Testing Automated Battery (CANTAB; Cantabeclipse, 2005) are from Lowe and Rabbit (1998); test-retest reliabilities for the pattern span task are from Della Sala, Baddeley, Gray, and Wilson (1997); and test-retest reliabilities for the trail-making task are from Spreen and Strauss (1991). For the remaining tasks, standardized alpha coefficients were calculated from the data of the present study.

All the tasks (except listening and counting span) were nonverbal. Four computerized tasks were drawn from the Cambridge Neuropsychological Testing Automated Battery (CANTAB; Cantabeclipse, 2005): spatial span, spatial working memory, rapid visual information processing, and intradimensional-extradimensional shift. Additional computerized tasks were programmed in E-Prime: change detection, counting span, and go/no go. Two of the remaining tasks were paper-and-pencil tasks (pattern span, trail making), and the final one (listening span) was administered orally. Responses on computerized tasks were measured by touch-screen, press-pad, or computer-key presses. All tasks began with practice trials to ensure that the children understood the instructions. A brief description of each task, along with reliability estimates, is provided in Table 2 (for a fuller description, see Rose, Feldman, & Jankowski, 2011).

Data analysis

Because the sample included preterm and full-term groups, all relations among measures were initially examined separately in each group. Given that the number of relations differing across the groups did not exceed what would be expected by chance, data from the two groups were combined (and correlations were partialed for group status to avoid inflation due to mean differences between groups).

For the age-11 assessments of EFs, the data were first evaluated for univariate and bivariate outliers (values > 2.5 SD from the mean or regression line); outlying values were removed (four data points from rapid visual processing, two from spatial working memory, and one from trail making). Then, a three-factor CFA was evaluated with SEM, using LISREL (Version 8.54; Jörskog & Sörbom, 2003) and maximum likelihood estimation. The model posited one factor each for Working Memory, Inhibition, and Shifting. Missing data (1.64%) were imputed using the Expected Maximization algorithm in PRELIS, and model fit was assessed using the normal-theory-weighted least squares χ², the root-mean-square error of approximation (RMSEA), and the comparative fit index (CFI). Values indicative of good fit are a nonsignificant χ² (p > .05), an RMSEA less than .05, and a CFI greater than .90. An alternative four-factor model, in which the simple and complex tasks were treated as separate working memory factors, was evaluated and compared with the three-factor model, using the χ² difference test.

SEM was used to evaluate two path models. In the first, a latent variable representing each EF was predicted from infant latent variables of Attention, Processing Speed, and Memory. In the second, latent variables from the toddler period were substituted for the infant ones.

Results

Confirmatory factor analysis of age-11 executive functions

In the CFA for the three-factor model of EFs, the three factors—Working Memory, Inhibition, and Shifting—were allowed to correlate, following Miyake et al. (2000). Correlations among the measures used in the CFA are shown in Table 3.

Table 3.

Correlations Among the Measures of Executive Functions

Measure	1	2	3	4	5	6	7	8	9
Working memory: storage
1. Spatial span	—
2. Pattern span	.30	—
3. Change detection	.19	.31	—
Working memory: storage and processing
4. Spatial working memory	.39	.13	.21	—
5. Counting span	.22	.10	.21	.24	—
6. Listening span	.40	.23	.23	.37	.39	—
Inhibition
7. Go/no-go task (no-go trials)	.17	.01	.35	.17	.10	.16	—
8. Rapid visual processing	.19	.22	.36	.13	.20	.10	.26	—
Shifting
9. Trail making	.23	.19	.31	.22	.28	.25	.04	.30	—
10. Intradimensional-extradimensional shift (reversal trials)	.22	.06	.18	.11	.17	.10	.01	.17	.36

Note: All correlations were partialed for birth status (premature vs. full-term). Correlations greater than .17 are significant, p < .05.

N = 124–130. All measures were scaled so that high scores indicate good performance.

The model provided an excellent fit, χ²(32) = 41.88, p = .11; RMSEA = .05; CFI = .96. The standardized factor loadings ranged from .43 to .88, all ps < .05 (see Table 4), and the factors were correlated, with rs ranging from .27 to .56. These results indicate that the EFs form distinct, though correlated, latent factors.

Table 4.

Standardized Factor Loadings From the Confirmatory Factor Analysis of the Age-11 Measures of Executive Functions

	Latent factor
Measure	Working Memory	Inhibition	Shifting
Spatial span	.63
Pattern span	.43
Change detection	.49
Spatial working memory	.57
Counting span	.62
Listening span	.58
Go/no-go task		.44
Rapid visual processing		.88
Trail making			.81
Intradimensional-extradimensional shift			.45

Note: All loadings shown here were significant, p < .001. All measures were scaled so that high scores indicate good performance. Correlations among the factors were as follows—Working Memory–Inhibition: r = .48, p < .001; Working Memory–Shifting: r = .56, p < .001; Inhibition– Shifting: r = .27, p < .05.

The alternative four-factor CFA, which separated simple and complex working memory tasks, also fit the data, χ²(29) = 36.36, p = .16; RMSEA = .04; CFI = .97. However, it did not fit better than the three-factor model, Δχ²(4) = 5.52. This being the case, the more parsimonious three-factor model was adopted.

Predicting age-11 executive functions from information processing in infancy

A path model predicting all three age-11 EFs from information processing in infancy is shown in Figure 1. This model assumed that all three information processing domains in infancy contributed to the variance in each age-11 EF. The latent variables for Working Memory, Shifting, and Inhibition were formed from the same scores that loaded on their respective factors in the CFA. These three EFs were allowed to correlate, as in the CFA. The latent variables from infancy (Attention, Speed, and Memory) were formed from measures that had been found to load on these factors previously (Rose et al., 2004; Rose et al., 2005b). This model fit the data well, χ²(138) = 154.51, p = .16; RMSEA = .03; CFI = .94.

Fig. 1.

Structural equation model showing the relation between information processing in infancy and age-11 executive functions. Latent variables are represented by ovals, and observed variables by rectangles. Single-headed arrows between latent variables represent paths from infant to age-11 constructs; double-headed arrows represent correlations among the age-11 executive functions. Significant paths and correlations involving latent variables are indicated by solid lines (p ≤ .05); nonsignificant ones are indicated by dashed lines (p > .05). Numbers on the arrows going from latent variables to observed variables are factor loadings (all significant, p < .05); the smaller arrows indicate residual variance. As suggested by modification indices, the model also contained correlations between the error terms for counting span and listening span (r = .22) and between the error terms for change detection and spatial span (r = –.27). Parameter estimates are shown for the completely standardized solution. CANTAB = Cambridge Neuropsychological Testing Automated Battery (Cantabeclipse, 2005); ID-ED shifts = intradimensional-extradimensional shifts.

Prediction of Working Memory

Infant memory and processing speed both had significant paths to age-11 Working Memory (β = 0.23, p < .05, and β = 0.43, p < .01). Thus, faster speed and better memory in the 1st year of life independently contributed to this EF over a 10-year span. Together, the three aspects of infant information processing accounted for 17% of the variance in age-11 Working Memory.

Prediction of Shifting

Infant processing speed also significantly predicted age-11 Shifting (β = 0.33, p < .01), a result consistent with the key assumption that faster speed of processing underlies faster performance on many measures of EF. The three latent factors from infancy, taken together, accounted for 19% of the variance in this EF.

Prediction of Inhibition

Although none of the separate paths from infancy to age-11 Inhibition were significant, infant attention and processing speed were marginally related to age-11 Inhibition (p < .10). Taken together, the variables from infancy accounted for 9% of the variance in age-11 Inhibition.

Predicting age-11 executive functions from information processing in toddlerhood

Figure 2 presents the model predicting the same three EFs from the latent factors for information processing in toddlerhood. In this model, Attention and Speed were allowed to correlate inasmuch as the solution would not converge otherwise. As the figure shows, virtually the same pattern of relations was obtained when variables measured in toddlerhood were substituted for those measured in infancy. This model provided an excellent fit to the data, χ²(156) = 157.06, p = .46; RMSEA = .01; CFI = .98.

Fig. 2.

Structural equation model showing the relation between information processing in toddlerhood and age-11 executive functions. Latent variables are represented by ovals, and observed variables by rectangles. Single-headed arrows between latent variables represent paths from toddler to age-11 constructs; double-headed arrows represent correlations among the age-11 executive functions. Significant paths and correlations involving latent variables are indicated by solid lines (p ≤ .05); nonsignificant ones are indicated by dashed lines (p > .05). Numbers on the arrows going from latent variables to observed variables are factor loadings (all significant, p < .05); the smaller arrows indicate residual variance. As suggested by modification indices, the model also contained a correlation between the error terms for 24- and 36-month recall memory (r = .42). Parameter estimates are shown for the completely standardized solution. CANTAB = Cambridge Neuropsychological Testing Automated Battery (Cantabeclipse, 2005); ID-ED shifts = intradimensional-extradimensional shifts.

Prediction of Working Memory

Memory and processing speed in toddlerhood both had significant paths to age-11 Working Memory (β = 0.30 and β = 0.24, both ps < .05), as was the case with the infant measures. Thus, faster speed and better memory in toddlerhood independently contributed to this EF. Together, the three aspects of information processing in toddlerhood accounted for 16% of the variance in age-11 Working Memory.

Prediction of Shifting

Processing speed from the toddler period also significantly predicted age-11 Shifting (β = 0.27, p < .05), as did speed in infancy. Taken together, the variables for toddlerhood accounted for 8% of the variance in age-11 Shifting.

Prediction of Inhibition

None of the separate paths from toddlerhood to age-11 Inhibition were significant. Taken together, the toddler variables accounted for 5% of the variance in this age-11 EF.

Discussion

This study is the first to show that information processing abilities that are emerging in infancy and toddlerhood have long-term implications for EFs in preadolescence. The early assessments, obtained from a prospective longitudinal study, covered three domains—attention, processing speed, and memory. The same abilities were measured twice in infancy (7 and 12 months) and twice in toddlerhood (24 and 36 months). The age-11 assessments included a broad battery of measures covering three pivotal EF factors: Working Memory, Inhibition, and Shifting. SEM, with latent variables, was used in a path model predicting the three EFs from measures in infancy. There were two major findings. First, the model fit the data well, with measures of information processing in infancy accounting for 9% to 19% of the variance in Working Memory, Shifting, and Inhibition. Second, measures from two of the infant information processing domains—processing speed and memory—had independent paths to age-11 Working Memory (βs = 0.43 and .23), and processing speed had an independent path to age-11 Shifting (β = 0.33). Virtually the same pattern of relations between early information processing abilities and later EF factors was obtained when measures from toddlerhood were substituted for measures from infancy. Our findings extend the literature on cognitive development by suggesting that core abilities from infancy and toddlerhood relate longitudinally to later EFs.

What is novel, and striking, about the present findings is the long-term implications of these newly emerging infant abilities for EF skills many years later. These core abilities were previously found to be important for developmental outcomes in toddlerhood (Rose, Feldman, Jankowski, & Van Rossem, 2005; Rose et al., 2008). This study shows that their importance extends over even longer periods, and to other cognitive domains, namely, the EFs necessary for complex reasoning, problem solving, and planning. A further hallmark of the present study is that, unlike many previous EF studies, which have used single tasks to represent constructs, we used latent variables, which minimize error variance and reduce problems associated with task impurity.

Overall, these results add strong support to the contention that infant cognition provides the roots of later, more complex abilities. These findings are buttressed by work showing that measures capturing core aspects of information processing early in life were related to later IQ (Fagan, 2011) and by a recent study showing that self-restraint in toddlerhood was related to EFs in adolescence (Friedman, Miyake, Robinson, & Hewitt, 2011). In the latter study, developmental trajectories of self-restraint (based on data from ages 14, 20, 24, and 36 months) were related to EFs at age 17 years; less restraint was associated with lower scores on a Common EF factor (encompassing response inhibition) and with higher scores on the Shifting factor. What remains unclear is whether relations found between infancy and age 11 years are due to continuities in the basic abilities, with the basic abilities having a similar impact on EFs across development; to continuities in the EFs themselves; or to both. To resolve this question, researchers would need to devise ways to assess EFs or their precursors in infancy itself.

The present results also support the literature indicating that processing speed is a key factor underlying more complex abilities in older children and adults (Salthouse, 1996). For example, in models proposing a developmental cascade, age-related increases in processing speed are seen as resulting in age-related improvement in working memory, which in turn affects fluid reasoning (Kail, 2007). The present findings raise the possibility that such cascades begin much earlier than previously thought, at least with respect to the index of processing speed examined here, namely, RT. This possibility is consistent with two previous studies from our lab. In one, a cascade was found at age 11 years, with processing speed influencing all three EFs described by Miyake et al. (2000): working memory, inhibition, and shifting (Rose, Feldman, & Jankowski, 2011). In the other, cross-age continuities from infancy to age 11 were found for latent factors representing several basic abilities, including speed (Rose et al., 2012). Taken together, these findings, strongly suggest that the foundational role of processing speed in cascades leading to later EFs begins in infancy.

The present results also indicate that infant memory plays an important role in more complex abilities. Although the mediating effect of memory in EFs has not been examined in children and adults as extensively as processing speed has been studied, two studies indicate that memory, too, has a central role in more complex abilities. In one, the ease of retrieving information from long-term storage, a process involved in free-recall memory, was found to relate to performance on complex span tasks, which are commonly used to assess working memory (Unsworth & Engle, 2006). In the other, age-related declines in memory among older adults were associated with age-related declines in a number of different EFs (Salthouse, 2005).

In summary, infancy proves to be a period with burgeoning cognitive abilities that have substantial long-term implications, even for EFs assessed 10 years later, in preadolescence. Recent findings indicating that various aspects of attentional control can be improved in infancy through training (Wass, Porayska-Pomsta, & Johnson, 2011) raise the possibility of similarly improving other cognitive functions in infancy, in order to benefit later EFs. In fact, if the presence of EFs in infancy could be demonstrated, it might be possible to improve EFs directly during this period.

Footnotes

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Funding

This research was supported in part by Grants HD 13810 and HD 049494 from the National Institutes of Health.

Notes

References

Bachevalier

Brickson

Hagger

(1993). Limbic-dependent recognition memory in monkeys develops early in infancy. NeuroReport, 4, 77–80.

Cantabeclipse. (2005). Cambridge, England: Cambridge Cognition Ltd.

Cowan

(2011). The focus of attention as observed in visual working memory tasks: Making sense of competing claims. Neuropsychologia, 49, 1401–1406.

Damasio

(1994). Descartes’ error: Emotion, reason, and the human brain. New York, NY: Putnam.

Della Sala

Baddeley

Gray

Wilson

(1997). Visual Patterns Test (VPT). San Diego, CA: Harcourt.

Dougherty

T. M.

Haith

M. M.

(1997). Infant expectations and reaction time as predictors of childhood speed of processing and IQ. Developmental Psychology, 33, 146–155.

Fagan

J. F.

(2011). Intelligence in infancy. In Sternberg

R. J.

Kaufman

S. B.

(Eds.), The Cambridge handbook of intelligence (pp. 130–142). New York, NY: Cambridge University Press.

Fagan

J. F.

Holland

C. R.

Wheeler

(2007). The prediction, from infancy, of adult IQ and achievement. Intelligence, 35, 225–231.

Friedman

N. P.

Miyake

Corley

R. P.

Young

S. E.

DeFries

J. C.

Hewitt

J. K.

(2006). Not all executive functions are related to intelligence. Psychological Science, 17, 172–179.

10.

Friedman

N. P.

Miyake

Robinson

J. L.

Hewitt

J. K.

(2011). Developmental trajectories in toddlers’ self-restraint predict individual differences in executive functions 14 years later: A behavioral genetic analysis. Developmental Psychology, 47, 1410–1430. doi:10.1037/a0023750

11.

Gathercole

S. E.

Pickering

S. J.

Ambridge

Wearing

(2004). The structure of working memory from 4 to 15 years of age. Developmental Psychology, 40, 177–190.

12.

Hollingshead

A. B.

(1975). The four-factor index of social status. Unpublished manuscript, Yale University, New Haven, CT. (Available from A. B. Hollingshead, Department of Sociology, Yale University, New Haven, CT 06520)

13.

Jörskog

Sörbom

(2003). LISREL (Version 8.54) [Computer software]. Lincolnwood, IL: Scientific Software International.

14.

Jurado

M. B.

Rosselli

(2007). The elusive nature of executive functions: A review of our current understanding. Neuropsychology Review, 17, 213–233.

15.

Kail

(2007). Longitudinal evidence that increases in processing speed and working memory enhance children’s reasoning. Psychological Science, 18, 312–313.

16.

Kane

M. J.

Engle

R. W.

(2003). Working memory capacity and the control of attention: The contributions of goal neglect, response competition, and task set to Stroop interference. Journal of Experimental Psychology: General, 312, 47–70.

17.

Lowe

Rabbit

(1998). Test/re-test reliability of the CANTAB and ISPOCD neuropsychological batteries: Theoretical and practical issues. Neuropsychologia, 36, 915–923.

18.

McDonough

Mandler

J. M.

McKee

R. D.

Squire

L. R.

(1995). The deferred imitation task used as a nonverbal measure of declarative memory. Proceedings of the National Academy of Sciences, USA, 92, 7580–7584.

19.

McKee

R. D.

Squire

L. R.

(1993). On the development of declarative memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 397–404.

20.

Miyake

Friedman

N. P.

Emerson

M. J.

Witzki

A. H.

Howerter

(2000). The unity and diversity of executive functions and their contribution to complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology, 41, 49–100.

21.

Quamme

J. R.

Yonelinas

A. P.

Widaman

K. F.

Kroll

N. E. A.

Suave

M. J.

(2004). Recall and recognition in mild hypoxia: Using covariance structural modeling to test competing theories of explicit theories. Neuropsychologia, 42, 672–691.

22.

Rose

S. A.

Feldman

J. F.

(1995). Prediction of IQ and specific cognitive abilities at 11 years from infancy measures. Developmental Psychology, 31, 685–696.

23.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

(2001). Attention and recognition memory in the 1st year of life: A longitudinal study of preterm and full-term infants. Developmental Psychology, 37, 135–151.

24.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

(2004). Dimensions of cognition in infancy. Intelligence, 32, 245–262.

25.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

(2005a). Recall memory in the first three years of life: A longitudinal study of preterm and term children. Developmental Medicine & Child Neurology, 47, 653–659. doi:10.1017/S0012162205001349

26.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

(2005b). The structure of infant cognition at 1 year. Intelligence, 33, 231–250. doi:10.1016/j.intell.2004.11.002

27.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

(2009). Information processing in toddlers: Continuity from infancy and persistence of preterm deficits. Intelligence, 37, 311–320. doi:10.1016/j.intell.2009.02.002

28.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

(2011). Modeling a cascade of effects: The role of speed and executive functioning in preterm/full-term differences in academic achievement. Developmental Science, 14, 1161–1175. doi:10.1111/j.1467-7687.2011.01068.x

29.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

Van Rossem

(2005). Pathways from prematurity and infant abilities to later cognition. Child Development, 76, 1172–1184. doi:10.1111/j.1467-8624.2005.00843.x

30.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

Van Rossem

(2008). A cognitive cascade in infancy: Pathways from prematurity to later mental development. Intelligence, 36, 367–378. doi:10.1016/j.intell.2007.07.003

31.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

Van Rossem

(2011). The structure of memory in infants and toddlers: An SEM study with full-terms and preterms. Developmental Science, 14, 83–91. doi:10.1111/j.1467-7687.2010.00959.x

32.

Rose

S. A.

Feldman

J. F.

Jankowski

J. J.

Van Rossem

(2012). Information processing from infancy to 11 years: Continuities and prediction of IQ. Intelligence, 40, 445–457.

33.

Rose

S. A.

Tamis-Lemonda

C. S.

(1999). Visual information processing in infancy: Reflections on underlying mechanisms. In Balter

Tamis-Lemonda

C. S.

(Eds.), Child psychology: A handbook of contemporary issues (pp. 64–84). Philadelphia, PA: Psychology Press.

34.

Salthouse

T. A.

(1996). The processing-speed theory of adult age differences in cognition. Psychological Review, 103, 403–428.

35.

Salthouse

T. A.

(2005). Relations between cognitive abilities and measures of executive functioning. Neuropsychology, 19, 532–545.

36.

Salthouse

T. A.

(2011). What cognitive abilities are involved in trail-making performance? Intelligence, 39, 222–232.

37.

Sigman

Cohen

S. E.

Beckwith

Asarnow

Parmelee

A. H.

(1991). Continuity in cognitive abilities from infancy to 12 years of age. Cognitive Development, 6, 47–57.

38.

Spreen

Strauss

(1991). A compendium of neuropsychological tests: Administration, norms, and commentary. New York, NY: Oxford University Press.

39.

Squire

L. R.

Wixted

J. T.

Clark

R. E.

(2007). Recognition memory and the medial temporal lobe: A new perspective. Nature Reviews Neuroscience, 8, 872–883.

40.

Unsworth

Engle

R. W.

(2006). Simple and complex memory spans and their relation to fluid abilities: Evidence from list-length effects. Journal of Memory and Language, 54, 68–80.

41.

Wass

Porayska-Pomsta

Johnson

M. H.

(2011). Training attentional control in infancy. Current Biology, 21, 1543–1547. doi:10.1016/j.cub.2011.08.004