Limits...
Training-induced compensation versus magnification of individual differences in memory performance.

Lövdén M, Brehmer Y, Li SC, Lindenberger U - Front Hum Neurosci (2012)

Bottom Line: Initial mnemonic instructions reduced between-person differences in memory performance, whereas further practice after instruction magnified between-person differences.We conclude that strategy instruction compensates for inefficient processing among the initially less able.In contrast, continued practice magnifies ability-based between-person differences by uncovering individual differences in memory plasticity.

View Article: PubMed Central - PubMed

Affiliation: Center for Lifespan Psychology, Max Planck Institute for Human Development Berlin, Germany.

ABSTRACT
Do individuals with higher levels of task-relevant cognitive resources gain more from training, or do they gain less? For episodic memory, empirical evidence is mixed. Here, we revisit this issue by applying structural equation models for capturing individual differences in change to data from 108 participants aged 9-12, 20-25, and 65-78 years. Participants learned and practiced an imagery-based mnemonic to encode and retrieve words by location cues. Initial mnemonic instructions reduced between-person differences in memory performance, whereas further practice after instruction magnified between-person differences. We conclude that strategy instruction compensates for inefficient processing among the initially less able. In contrast, continued practice magnifies ability-based between-person differences by uncovering individual differences in memory plasticity.

No MeSH data available.


Related in: MedlinePlus

Graphical representation of the latent growth curve model implemented here. Observed variables are represented by squares, latent variables by circles, regression weights by one-headed arrows, and variances and covariances by two-headed arrows. The triangle indicates means. Unlabeled parameters are fixed to the values displayed in the matrix of loadings. IC, intercept, reflecting post-training performance; S, linear slope; S2, quadratic slope; SS1–7, session-wise linear slopes; l, list.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3351801&req=5

Figure 2: Graphical representation of the latent growth curve model implemented here. Observed variables are represented by squares, latent variables by circles, regression weights by one-headed arrows, and variances and covariances by two-headed arrows. The triangle indicates means. Unlabeled parameters are fixed to the values displayed in the matrix of loadings. IC, intercept, reflecting post-training performance; S, linear slope; S2, quadratic slope; SS1–7, session-wise linear slopes; l, list.

Mentions: We analyzed practice gains with a latent curve model (LCM; e.g., Bryk and Raudenbush, 1987; McArdle and Epstein, 1987; Meredith and Tisak, 1990; McArdle, 2006). Figure 2 displays a graphical representation of the LCM implemented here. The observed variables, l11–l52, emanate from the seven sessions in the phase of individually adaptive practice, each session including the presentation of six location-word lists. In a linear LCM, two latent variables, the intercept IC and the linear slope S, are proposed to account for the time series information. The linear slope S represents linear gain from practice by constraining the 42 loadings of the observed variables on S to increase linearly. The intercept IC represents an individual's latent score at the end of the time series (i.e., at l52) by setting the factor loading of the observed variable l52 on S to zero (i.e., l11 has a −41 loading on S, l12 has a −40 loading, etc.; see the loading matrix (Λ) in Figure 2). The intercept and the linear slope factors are estimated at the mean level (i.e., their means μIC and μS are estimated), they both allow for interindividual differences (i.e., their standard deviations σIC and σS are estimated), and they may covary ρIC, S. The error variance σ2e is commonly assumed to have a mean of zero and to neither correlate nor change over time. Estimating the six parameters mentioned so far (μIC, μS, σIC, σS, ρIC, S, σ2e) corresponds to estimating a classic linear LCM. We included an additional factor representing the orthogonal quadratic effect (S2). For these factors, preliminary analyses showed no significant interindividual differences (i.e., standard deviations) for any of the age groups. Therefore, we did only estimate the mean μS2 and not the standard deviation.


Training-induced compensation versus magnification of individual differences in memory performance.

Lövdén M, Brehmer Y, Li SC, Lindenberger U - Front Hum Neurosci (2012)

Graphical representation of the latent growth curve model implemented here. Observed variables are represented by squares, latent variables by circles, regression weights by one-headed arrows, and variances and covariances by two-headed arrows. The triangle indicates means. Unlabeled parameters are fixed to the values displayed in the matrix of loadings. IC, intercept, reflecting post-training performance; S, linear slope; S2, quadratic slope; SS1–7, session-wise linear slopes; l, list.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3351801&req=5

Figure 2: Graphical representation of the latent growth curve model implemented here. Observed variables are represented by squares, latent variables by circles, regression weights by one-headed arrows, and variances and covariances by two-headed arrows. The triangle indicates means. Unlabeled parameters are fixed to the values displayed in the matrix of loadings. IC, intercept, reflecting post-training performance; S, linear slope; S2, quadratic slope; SS1–7, session-wise linear slopes; l, list.
Mentions: We analyzed practice gains with a latent curve model (LCM; e.g., Bryk and Raudenbush, 1987; McArdle and Epstein, 1987; Meredith and Tisak, 1990; McArdle, 2006). Figure 2 displays a graphical representation of the LCM implemented here. The observed variables, l11–l52, emanate from the seven sessions in the phase of individually adaptive practice, each session including the presentation of six location-word lists. In a linear LCM, two latent variables, the intercept IC and the linear slope S, are proposed to account for the time series information. The linear slope S represents linear gain from practice by constraining the 42 loadings of the observed variables on S to increase linearly. The intercept IC represents an individual's latent score at the end of the time series (i.e., at l52) by setting the factor loading of the observed variable l52 on S to zero (i.e., l11 has a −41 loading on S, l12 has a −40 loading, etc.; see the loading matrix (Λ) in Figure 2). The intercept and the linear slope factors are estimated at the mean level (i.e., their means μIC and μS are estimated), they both allow for interindividual differences (i.e., their standard deviations σIC and σS are estimated), and they may covary ρIC, S. The error variance σ2e is commonly assumed to have a mean of zero and to neither correlate nor change over time. Estimating the six parameters mentioned so far (μIC, μS, σIC, σS, ρIC, S, σ2e) corresponds to estimating a classic linear LCM. We included an additional factor representing the orthogonal quadratic effect (S2). For these factors, preliminary analyses showed no significant interindividual differences (i.e., standard deviations) for any of the age groups. Therefore, we did only estimate the mean μS2 and not the standard deviation.

Bottom Line: Initial mnemonic instructions reduced between-person differences in memory performance, whereas further practice after instruction magnified between-person differences.We conclude that strategy instruction compensates for inefficient processing among the initially less able.In contrast, continued practice magnifies ability-based between-person differences by uncovering individual differences in memory plasticity.

View Article: PubMed Central - PubMed

Affiliation: Center for Lifespan Psychology, Max Planck Institute for Human Development Berlin, Germany.

ABSTRACT
Do individuals with higher levels of task-relevant cognitive resources gain more from training, or do they gain less? For episodic memory, empirical evidence is mixed. Here, we revisit this issue by applying structural equation models for capturing individual differences in change to data from 108 participants aged 9-12, 20-25, and 65-78 years. Participants learned and practiced an imagery-based mnemonic to encode and retrieve words by location cues. Initial mnemonic instructions reduced between-person differences in memory performance, whereas further practice after instruction magnified between-person differences. We conclude that strategy instruction compensates for inefficient processing among the initially less able. In contrast, continued practice magnifies ability-based between-person differences by uncovering individual differences in memory plasticity.

No MeSH data available.


Related in: MedlinePlus