Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

Rev. Neurosci.

2017; aop

Carlos M. Gómez*, Catarina Isabel Barriga-Paulino, Elena Isabel Rodríguez-Martínez,


Ma Ángeles Rojas-Benjumea, Antonio Arjona and Jaime Gómez-González

The neurophysiology of working memory


development: from childhood to adolescence
and young adulthood
https://doi.org/10.1515/revneuro-2017-0073
Received August 23, 2017; accepted September 9, 2017
Introduction
Abstract: Working memory (WM) is an important cog- The concept of working memory (WM) refers to the brain
nitive function that is necessary to perform our daily system involved in temporal storage and manipulation of
activities. The present review briefly describes the most information. This mechanism makes it possible to perform
accepted models underlying WM and the neural networks complex cognitive functions (such as language, compre-
involved in its processing. The review focuses on how the hension, learning, and reasoning) (Baddeley, 1992). WM
neurophysiological mechanisms develop with age in the corresponds to a system capable of maintaining a limited
periods from childhood to adolescence and young adult- amount of information active for a short period of time (in
hood. Studies using behavioral, neuroimaging, and elec- terms of seconds) and process this information.
trophysiological techniques showed the progress of WM In this review, we focus on the neurophysiology of WM
throughout the development. The present review focuses during development from childhood to adulthood (from
on the neurophysiology of the basic processes underlying 4–5 to 24–25  years old, approximately). This approach
WM operations, as indicated by electroencephalogram- complements excellent reviews about behavioral, struc-
derived signals, in order to take advantage of the excellent tural, and blood-flow studies related to WM development
time resolution of this technique. Children and adults use (Cowan, 2010; Simmering and Perone, 2013; Constanti-
similar cerebral mechanisms and areas to encode, recog- nidis and Klingberg, 2016; Klingberg, 2016), and other
nize, and keep the stimuli in memory and update the WM reviews that concentrate on WM and more tangentially
contents, although adults rely more on anterior sites. The address the problem of WM development (Baddeley, 2012;
possibility that a functional reorganization of WM brain Luck and Vogel, 2013; D’Esposito and Postle, 2015; Eriks-
processing occurs around the adolescent period is sug- son et al., 2015; Myers et al., 2017). The neurophysiology
gested, and would partly justify the high prevalence of the of infants and toddlers’ WM development (see Reynolds
emergence of mental pathology in the adolescent period. and Romano, 2016, for a review) is not covered by itself,
but as previous stages of development. The behavioral,
Keywords: brain oscillations; cerebral maturation; devel-
cognitive, neuroimaging, and theoretical comments in the
opment; event-related potentials; working memory.
present review are limited to allowing an integration of the
neurophysiology of WM and the behavioral/neuroimaging
windows of WM operations during the development. The
electroencephalographical (EEG) and magnetoencepha-
*Corresponding author: Carlos M. Gómez, Laboratory of Human lographical (MEG) studies expand our knowledge of WM
Psychobiology, Department of Experimental Psychology, University development due to the excellent temporal resolution
of Seville, Calle Camilo José Cela, S/N, E-41018 Sevilla, Spain, offered by these techniques. In this regard, neuroimag-
e-mail: cgomez@us.es
ing techniques have difficulties in separating the differ-
Catarina Isabel Barriga-Paulino: Laboratory of Human
Psychobiology, Department of Experimental Psychology, University ent processing stages of WM operation, such as encoding,
of Seville, E-41018 Sevilla, Spain; and Center for Biomedical maintenance, recognition, or updating, which occur in
Research, Universidade do Algarve, 8005-140 Faro, Portugal the range of few seconds and subseconds, whereas the
Elena Isabel Rodríguez-Martínez, Ma Ángeles Rojas-Benjumea and time resolution windows of MEG and EEG can account for
Antonio Arjona: Laboratory of Human Psychobiology, Department
these fast processes.
of Experimental Psychology, University of Seville, E-41018 Sevilla,
Spain
The broad conceptualization of this central cogni-
Jaime Gómez-González: UGC de Salud Mental, Hospital Virgen tive process makes it difficult to review all the possible
Macarena, Servicio Andaluz de Salud, E-41009 Sevilla, Spain instances. In order to systematize the developmental

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
2      C.M. Gómez et al.: Working memory development

issues, the discussion is organized by multicomponent, visual appearance (Logie, 1995). More recently, the epi-
constructive operators, and state-based models. In addi- sodic buffer was added as an element of interconnection
tion, the present review is limited to the periods of child- between the other elements. The inclusion of this compo-
hood, adolescence, and young adulthood. nent stems from new data indicating that phonological
and visual information are combined, and it also inte-
grates the information that comes from the LTM, creating
a multimodal and temporally ordered representation of
WM models – their relevance the current situation (Baddeley, 2012).

for WM development
Constructive operator theory
Multicomponent model
An alternative structural theory of WM to the one pre-
Since the emergence of this construct, several WM theo- sented by Baddeley and Hitch (1974), which was devel-
ries have been proposed. One of the most extensively oped to respond to the ontogenetic evolution of WM,
studied, which describes its operations and structure, corresponds to Pascual-Leone (Pascual-Leone and Bail-
is the WM model proposed by Baddeley and Hitch (1974; largeon, 1994) and includes schemas as central processing
recently reviewed in Baddeley, 2012). In this model, elements (figurative, more related to content, and opera-
the visual and verbal modalities are explicitly distin- tive, more related to processing) (reviewed in Kemps et al.,
guished. WM is composed of a main system called the 2000). This theory also includes the so-called hardware
central executive, which is assisted by two secondary operators, which would correspond to certain primitive
systems: the phonological loop and the visuo-spatial processing, transversal to the different schemes. From the
sketchpad. More recently, the so-called episodic buffer, standpoint of development, the most important capacity
which would make it possible to integrate the infor- would be the operational memory, which would increase
mation from the two slave systems and the long-term by one information unit every 2 years, reaching the capac-
memory (LTM), has been included in the model. Most ity of seven at the age of 15.
of the evidence for this model comes from interference According to the constructive operator theory, cog-
studies, where any of the model’s compartments can be nitive performance is co-determined by the interaction
interfered with independently by concurrent demand- between schemas and hardware operators. When an input
ing tasks (Baddeley, 2012). is given, a series of schemas are activated, and together
The central executive is defined as a system of high- they constitute the ‘field of mental attention’ or WM.
level attention control required in novel situations such The selection and activation of these schemas depends
as problem solving, planning, or strategy selection. The on three mechanisms: the M-operator, the I-operator (an
central executive would be able to focus and divide the operator of inhibition of the non-relevant schemas), and
attention in order to have control over the current pro- the execution schemes (although Pascual-Leone’s opera-
cesses and interface with the LTM. This component is in tor theory is considerably more complex and includes up
charge of coordinating the attention during tasks with to 10 operators). The idea of the M-operator, related to the
immediate processing demands, and coordinating infor- number of schemas that can be held simultaneously in
mation between the visual and verbal systems. It is also the mental attention, presents a certain theoretical reso-
important on complex tasks that require simultaneous nance to WM models that conceptualize WM as an atten-
multi-task processing. tionally activated element of the LTM (Cowan, 1995, see
The phonological loop is responsible for the tempo- below).
rary storage and processing of verbal and auditory infor-
mation, which can be sustained in the form of memory
signals that spontaneously disappear (in a time frame of State-based models
2–3 s), unless it is updated (Baddeley, 1996, 1998).
The visuo-spatial sketchpad is the system responsi- Other alternative WM models are the ‘state-based WM
ble for the temporal retention and manipulation of visual models.’ These models consider WM as the activation of
and spatial information (Logie et al., 1990). This compo- semantic, perceptual, or motor LTM representations. The
nent is capable of supporting spatial information, such information stored in the LTM is maintained active, at dif-
as the location of an object, and information about its ferent levels of intensity, throughout the WM operation.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      3

In these types of models, attention plays a central role Neural networks of WM in DRTs
as a determinant of which elements are activated in
WM, and at what intensities. Cowan (1995) and other and DMTS tasks
authors (McElree, 2006; Oberauer, 2013) have progres-
DRTs, introduced by W.S. Hunter at the beginning of the
sively refined the theory and propose that certain items
20th century, are frequently used in the study of WM. In
can be temporarily activated in the LTM, basically in two
these tasks, sensory stimulation and motor responses are
forms: one directly controlled by the focus of attention
separated by a brief period during which sensory infor-
(FoA) (with limited capacity), and the other activated in
mation must be kept in memory in order to produce the
a portion of the LTM (activated LTM) (with much larger
correct response in the test phase. Several injury studies
capacity). Once attention is not focused on some of the
revealed that this type of task was a good paradigm to
items activated in the FoA, they are transferred to the
study the prefrontal functions because deficits in perform-
activated LTM, where they remain in a latent active state
ing the test were observed in animal subjects with damage
and can be easily reintroduced into the FoA. This type
in the prefrontal region (e.g. Goldman-Rakic, 1971).
of model can be associated with the so-called ‘sensori­
In the early 1970s, the first recordings were made by
motor recruitment,’ where the activated elements are the
sensory or motor networks that represent the perceptual Fuster and Alexander (1971) and Goldman-Rakic (1995)
object or the motor action (D’Esposito and Postle, 2015). on neurons of awake monkeys. Using this electrophysi-
Attention would play the role of maintaining activated ological technique in a DRT, these authors discovered
the representations of recently presented objects that ‘memory cells’ in the prefrontal and temporal cortex that
are important for the completion of the task. Interest- had high discharge rates during the delay period while the
ingly, representations can also be prospectively activated animal maintained sensory information in the absence of
in attentional cueing paradigms (where a cue indicates stimulation.
the location of the next stimulus). This prospective acti- The ‘recognition memory’ task, also called the DMTS
vation can be observed behaviorally through a reduction task, consists of memorizing a stimulus sample at the
in response times in the cued location, but also through beginning of the trial; after a delay period, the same stim-
the presence of the contingent negative variation (CNV) ulus is presented again in conjunction with other distinct
(Gómez et al., 2007; Gómez and Flores, 2011), suggesting stimuli (Fuster and Jervey, 1981). The aim of the task is
that (as proposed by theorists of the state-based models) to recognize and identify memorized stimuli (Löw et al.,
attention could be a mechanism for keeping elements in 1999). In the trained animal, the sample stimulus (in this
the WM activated. These types of state-based models also case visual), which appears at the beginning of each trial,
resemble Fuster’s proposal (2007) for WM mechanisms, activates an extensive network encompassing the neural
where WM would correspond to the activation of sensory representations of perception and action associated with
and executive networks. In the same vein, using retrocues the stimulus; both perceptual and motor memories are
(which are presented after the stimulus set presentation activated. In this way, the stimulus excites cells in the
and indicate the relevant features of the items stored in posterior cortex, involved in the processing of the sample
the WM for future action), Myers et  al. (2017) proposed stimulus to be retained, and in frontal areas, related to
that internal attention, driven by the retrocue, would act the processing of the motor response. An example was
in a similar manner to the biased competition model of observed by Fuster and Jervey (1982) on a task where the
external attention, generating a prioritization of the ret- monkey had to squeeze a lever of a similar color to the one
rocued item to be recovered. previously presented. These authors observed a neural
Baddeley (2012) reported that the difference between activation in the inferotemporal lobe during the delay
the multicomponent model and the state-based models period, which could reach 15 s in duration, suggesting that
(as proposed by Cowan, 1995) lies more in the terminol- these neurons are participating in the memory of the pre-
ogy than in fundamental processes. Thus, the control of viously presented visual stimulus. Given that the monkey
the episodic buffer by the central executive would not be had to retain this stimulus during the delay period and
different from the attentional focus on the LTM. At this then respond correctly, the network representing the stim-
point, it is critical to understand how the brain is able to ulus has to remain active during that period, and, there-
activate neural representations, and two related tasks that fore, the inferotemporal memory cells are active during
have contributed to clarifying this topic are the delayed the retention period. In addition, because the stimulus is
response task (DRT) and the delayed match-to-sample an indicator for future action, it selects a motor memory
(DMTS) task. that activates the prefrontal network while it is being held

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
4      C.M. Gómez et al.: Working memory development

in memory (Fuster, 1997). Therefore, the shared activation Behavioral development of WM


observed in the inferotemporal and prefrontal neurons,
using the visual DMTS, reflects the interactions within the Although there is a large amount of literature showing an
vast memory network of the frontal and posterior cortex increase in WM performance with age (Luciana et al., 2005;
in the visual WM. Thus, it can be postulated that, on this Conklin et al., 2007; Tamnes et al., 2013; Pelegrina et al.,
type of task, the memorandum consisting of a stimulus 2015), the present review is focused on the type of WM
to be retained for future action causes the activation of models described in the previous section, and more specif-
a large network with two main components: an infero­ ically, WM neurophysiological development during child-
temporal network representing the visual memo and a hood, adolescence, and young adulthood, rather than in
prefrontal network prospectively representing the motor infancy. The interest in this age stems from the fact that
response at the end of each trial (Fuster, 1999; Fuster and during the transition from childhood to young adulthood,
Bressler, 2014). there is a second burst of brain and behavioral changes that
A couple of points should be added to the are important not only for the maturational understanding
neurophysiology of the DMTS: (i) the perceptual neural of WM, but also to understand and prevent psychopatholo-
representation of the items to be maintained in memory gies that frequently arise at these ages (Paus et al., 2008).
would be distributed in cortices specialized in processing The development of WM can be analyzed through behavio-
these types of stimulus (visual, auditory, somatosensory, ral parameters (obtained from reaction times on the DMTS
etc.); (ii) the mechanisms for maintaining the items in and other WM tasks as the N-back) or through the scores
WM are still disputed; they could be in the reverberating obtained on the Working Memory Test Battery for Children
activity between prefrontal and specialized cortices, as (WMTB-C), developed by Pickering and Gathercole (2001)
suggested by the impairment in WM operation after cooling to operationalize the Baddeley and Hitch model, as well
the prefrontal cortex (Fuster, 1997) and in which brain as many other psychometric tests. The Pasqual-Leone WM
oscillations would be a determinant mechanism (Bressler model has been studied through the Mr. Peanut test, built
and Richter, 2015), or they could be based in molecular-­ by DeAvila (1974) and Diaz (1974).
synaptic changes that maintain persistent neural activities WM plays a very important role in the development
in the network (reviewed in D’Esposito and Postle, 2015). of cognition in the child. With regard to the multicom-
The idea of the synaptic changes related to experience ponent model of WM, studies in children from 5–6 to
being ubiquitous in the brain would provide support for 11–12  years old showed that the greatest increase in the
the state-based models of WM. efficacy of this cognitive function occurs after 7 years of
Human neuroimaging studies have extensively ana- age (Hitch et al., 1989). Gathercole (1998) states that, after
lyzed neural activation during WM tasks. Frontal, cin- age 7, the WM resembles that of adults in the organiza-
gulate, parietal, visual, and cerebellar areas have been tion and strategies used. The way to code the retained
described as activated during WM tasks (Cabeza and information changes during development (Hitch et  al.,
Nyberg, 2000). The possibility has been proposed that the 1989; Miles et al., 1996). Young children use the specific
processes related to the maintenance of information are characteristics of objects, such as shape, orientation, and
based on frontal networks and broadly distributed areas arrangement of details, but after 8 years of age, the use of
related to the content to be maintained. This notion, pre- phonological verbal coding is implemented. This mecha-
viously suggested by Fuster (2007), has been extended nism could be an important contributor to the higher effi-
to humans by studying the effects of transcranial mag- cacy of visuo-spatial WM as age increases. Thus, a crucial
netic stimulation in control subjects and in patients with period of developmental change occurs around 7  years
frontal strokes, who lose the capacity to focus attention old. Children under 7  years old rely more on the visual
and select the maintained items (D’Esposito and Postle, store of short-term memory (to remember materials such
2015). However, by using multi-voxel pattern analysis, as family photos and nameable objects), and from the age
the idea that content information representation during of 7 onwards, children prefer to rely on the phonological
WM tasks could be more distributed in the cortex (includ- loop, by recoding the pictorial forms into the associated
ing frontal areas) was recently proposed (Lee and Baker, spoken names. The increase in the rehearsal rate would
2016). By contrast, a recent review (Leavitt et  al., 2017) facilitate the increase in the capacity in the phonological
examining the electrophysiological and BOLD signals loop and in the verbally recoded visuo-spatial sketchpad
across the cortex during WM tasks suggested the prefer- (Gathercole et al., 2004).
ence for sustained activity to be implemented in prefron- The stimuli used in the visuo-spatial tests of the
tal and sensory association cortices. WMTB-C are selected on the basis of being very difficult

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      5

to recodify verbally because they do not correspond to In a confirmatory factorial analysis study with subjects
namable family objects. On tasks involving such materi- ranging from 6 to 15  years old, Gathercole et  al. (2004)
als, older children and adults maintain confidence in non- found that the structure of a three-multicomponent model
verbal forms of the short-term memory store. The current is already present in the youngest group, which is consist-
understanding about the development of the central ent with the different neural networks supporting each of
executive and the non-verbally coded visuo-spatial these components.
memory is less advanced than the knowledge about the The maturational growth in M-capacity, defined as the
development of the phonological loop (Pickering, 2001). maximum number of schemes that can be simultaneously
The consistent findings from many studies show that per- activated by attentional resources (Pascual-Leone and
formance on WM tests increases steadily throughout the Baillargeon, 1994), accounts for evolutionary changes in
childhood years, with performance at 15 years old nearing cognitive performance, such as language (Johnson et al.,
the adult level, and the central executive component 1989), motor performance (Todor, 1979), and moral judg-
reaching maturation later than the other components. It ment (Stewart and Pascual-Leone, 1992). The prediction
has been proposed that the processing efficiency of the of a validated quantitative increase in the WM capacity,
central executive improves with age, which would allow related to age, has been empirically demonstrated. Accord-
the release of space for the storage of memory items (Case ing to Pascual-Leone and Baillargeon (1994), cognitive
et  al., 1982). This improvement in the central executive development is determined by a gradual increase in one
would facilitate the increased performance in the slave unit of M-capacity every 2  years, from 3  years of age to
sub-components. However, particularly on complex tasks late adolescence. This increase in capacity is one of the
that take a long time to complete, an increase in the time distinctive aspects of the constructive operator theory.
that a particular item can be maintained in the WM would Some studies have suggested that the constructive operator
be another factor explaining the improvement with age. theory, which considers the M-operator as central to WM
This latter aspect can be studied through the DMTS tasks, operations, and the multicomponent model of WM, may be
which make it possible to control the time duration of the complementary (Kemps et al. 2000). The ­M-operator would
delay period. account more parsimoniously for the increase in capacity,
Because WM development studies have focused whereas the multicomponent model would more easily
largely on the changes taking place within the individual accommodate the structure of WM. In fact, the increase in
components of the model, relatively little is known about capacity with age, the interference in spatial information
the internal structure of the WM system in general, and coded on the Mr. Peanut test by articulatory concurrent
whether it changes with age. A natural way to study this tasks in older children, and the interference in Mr. Peanut
particular aspect is to obtain the pattern of performance task performance by a concurrent Corsi Block task at any
covariation in the different components of the model as age suggest: (i) an increase in available WM space with age
age progresses. A small number of studies have inves- (M-operator) and (ii) the existence of two slave systems
tigated the relationships between WM components in that can interact through the verbal recoding of spatial
children. The data provided by Pickering et  al. (1998) information in the phonological loop (multicomponent
indicate that between the ages of 5 and 8, the phonologi- model) around the age of 8 (Kemps et al. 2000).
cal loop and visuo-spatial sketchpad are independent DMTS tasks are closely related to the state-based
from one another. In a study of 6- and 7-year-old children, models, for instance, in the WM model proposal by Fuster
Gathercole and Pickering (2000) reported evidence that (2007), where the transient activation of long-term repre-
the central executive and the phonological loop were sentations of items in different sensory cortices is central
separable, but moderately associated with each other, to understanding the representational role of WM. With
which is consistent with the adult WM model. Short-term regard to the study of WM development using DMTS
visual-spatial memory, on the other hand, was not dis- tasks, the issue about the evolution of the encoding,
sociable from the central executive function, suggesting retention, and matching (recognition) phases has hardly
that it cannot represent an independent entity (at least been addressed (Paule et al., 1998). The questions about
at this point in development) (see also Wilson et  al., the changes with age have mostly been related to: (i) how
1987). Jarvis and Gathercole (2003) evaluated children many items can be held in memory (WM span) and (ii)
between 11 and 14  years of age on complex verbal and how the delay period affects performance (transient atten-
visual-spatial memory capacity, and, at both ages, these tion span).
two aspects (whether based on complex capacity or only Luciana et  al. (2005) analyzed non-verbal tasks,
on storage measures) were independent from each other. including a facial recognition memory test and a spatial

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
6      C.M. Gómez et al.: Working memory development

delayed response test, in subjects aged 9 to 20. The authors between 7 and 9 years of age. The rate of decay in preci-
observed that on the visual recognition test, the perfor- sion, as the delay time increased, was more pronounced
mance with respect to precision (percentage of correct for younger children, indicating worse retention of the
responses) did not change with age, indicating that rec- stimulus. In 4-year-old children, precision decreased
ognition memory for non-verbal (visual) stimuli seems to by 20% between the shortest delay period (1  s) and the
mature early, before the age of 9. On more complex tasks, longest delay period (32  s). Instead, for the 11-year-old
such as the spatial delayed response test, which also children, during the same delay periods, the decline was
includes the recall operation (remembering where the dot only 3%. Older children were more accurate at recogniz-
target was displayed), the performance in the group of ing the correct stimulus in the shortest delay period (1 s)
older subjects (18–20 years) was significantly better than compared to younger children, indicating better encoding
in the other age groups (9–10, 11–12, and 13–15  years), of the stimulus. These data indicate that there are differ-
but it was not different from the group of 16–17 year olds. ences in encoding and retention of stimuli in children of
The latter group performed better than the 9- to 10-year- different ages; more specifically, younger children appear
old and 11- to 12-year-old groups, but it was only slightly to have more difficulty with encoding and retaining
better than the 13- to 15-year-old group. For the 9–10, 11–12, information.
and 13–15 age groups, there were no differences in perfor- Closely related to WM studies with DMTS tasks are
mance. Reaction times did not vary between groups when those that fall under the umbrella of the visual WM. In the
there was no delay period between stimuli. The same latter concept, the focus is on the retention process rather
study also showed that the spatial memory span contin- than in WM items processing. In the change detection
ues to improve until adolescence. This study suggests ado- paradigm, the subject must recognize the change of one
lescence as an end point for maturation of WM in DMTS item in an array. This type of paradigm finds a capacity
tasks. saturation at around 3 or 4 items (Luck and Vogel, 2013).
Regarding the problem of the transient attention Ross-Sheehy et al. (2010) showed that this limit capacity
span, in this same study (Luciana et al., 2005), when two would be already in place at the age of 10 months. A dis-
interval levels between the stimuli (500 and 8000  ms) cussion of the continuous (resource-sharing models) and
were analyzed, the authors observed that the youngest slots models (available discrete WM slots) for WM attained
group of subjects (9–10 years old) was significantly less by the use of this paradigm is not relevant to the present
accurate in the longer interval (8000  ms) compared to review (see Luck and Vogel, 2013).
the other four age groups (which showed no differences It can be concluded that performance on WM tasks
between them), suggesting that the ability to sustain the improves with age (capacity, speed, retention time and
target position for long periods of time does not show accuracy). Children’s rate of improvement (increase
developmental changes after the age of 11–12. Therefore, in accuracy and decrease in reaction times) is greater
this study suggests that recall and transient attention at younger ages (Chelonis et  al., 2000). However, as
span operations mature later than the recognition function. Simmering and Perone (2013) indicated, the huge
­
The ability to increase the time that items can be held variability across tasks and modalities during development
in WM with age would be explained by the time-based generates difficulties in interpreting the processes underlying
resource-sharing model, which proposes that attention WM. In particular, the variability in WM capacity results
must switch from the maintenance to the processing on different tasks is illustrated in tables 1–4 by S­ immering
processes in WM. The increase in processing efficiency and Perone (2013). In the opinion of these authors, rather
with age would allow more frequent engagement of attention than regarding capacity as a constitutive factor of WM, it
toward the maintenance operation, thus increasing the should be viewed as the result of the interaction between
time that items are held in the WM (­Barrouillet et  al., dynamic systems at the behavioral and cognitive levels
2009). that change during development, with capacity changing
Paule et  al. (1998) carried out a study with children as a by-product of the interaction. In any case, the increase
between 4 and 12 years old using a DMTS-type paradigm in capacity and maintenance time with age remains an
with six different waiting periods (between 1 and 32 s), in important characteristic of WM development (Gathercole
order to determine how normal development affects the et  al., 2004; Barrouillet et  al., 2009; Cowan, 2010). The
latency of the response, the encoding, and the stimulus interesting question of how changing higher-order strate-
retention. The results showed that response accuracy gies in WM operations (e.g. chunking) would be related to
increased as the children grew, whereas the variability the increase in WM performance with age is beyond the
decreased with age, with the largest decreases occurring scope of the present review.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      7

Neuroimaging studies of WM the online maintenance of the object). Using near infra-
red spectroscopy during a spatial WM task, an increase
development in activity in the dorsolateral prefrontal cortex from 3 to
7 years old has also been observed (Perlman et al., 2016).
The relationship between anatomical and physiological
Taken together, previous results suggest that the fronto-
brain maturation and WM development has been ana-
parietal-temporal networks, which are crucial for WM
lyzed through magnetic resonance studies. Neuroimaging
operation in adults, begin to be recruited early in child-
studies using functional magnetic resonance have shown
hood (from the age of at least 6).
that similar brain structures are involved in children
WM behavioral improvement with age has also been
and adults during WM operations, but children present
associated with neuroanatomical changes. The increase
a lower level of activation (Kwon et al., 2002; Klingberg,
in structural connectivity between frontal and parietal
2006). Increased activity in lateral areas of the prefrontal
areas (Ostby et  al., 2011; Vestergaard et  al., 2011) and
cortex (Brodmann areas 9, 46) was observed during a
inside frontal areas (Zhou et  al., 2014) has been related
visuo-­spatial memory recall task in adults and 5- to 6-year-
to better WM performance. The increase of myelin in the
old children (Tsujimoto et  al., 2004), indicating that the
superior longitudinal fasciculus and the cortical thin-
areas and characteristics of this activity are similar in
ning in fronto-parietal areas (probably due to synaptic
adults and pre-school children. These authors have dem- pruning) are independent predictors of WM performance
onstrated that the lateral prefrontal cortex of pre-school with age (Ostby et al., 2011).
children is active during the WM process, revealing that, Results seem to agree about the use of similar struc-
at 5–6 years of age, this brain structure has already devel- tures in children and adults for WM computations.
oped the processing of this important cognitive function. However, this view must be reconciled with the fact that,
The performance of 8- to 10-year-old children on a during development, different structures could be partici-
spatial N-back task activated the same regions observed pating in WM. There is some evidence that children would
in adults, including the right dorsolateral prefrontal and use some different structures from adults to perform the
parietal cortices (Thomas et  al., 1999). Using the same WM functions. Goldman-Rakic (1971) injured the dor-
type of test (a visuo-spatial N-2 back) in subjects between solateral frontal cortex of monkeys in infant and adult-
the ages of 7 and 22, Kwon et al. (2002) observed that the juvenile ages and subjected them to a DRT task. She found
degree of activation of some areas in the prefrontal and that small monkeys did the job well, and, interestingly,
parietal cortex of both hemispheres increases with age. monkeys injured in adulthood performed the task with
In this study, although the task performed was visuo- low accuracy. This implies that, in small monkeys, the WM
spatial, the increase in activation in the left hemisphere task is performed in structures that are not exactly similar
was observed in areas related to the phonological loop, in to the dorsolateral frontal cortex. In addition, when the
particular the Broca area. The authors argued that this is injured infant monkeys grew, they performed the task
due to the development of a strategy that allows the verbal worse than in the infantile stage, probably because the
coding of visual information at the age of 7. These observa- transfer of the function to the dorsolateral frontal cortex
tions revealed that visual and verbal processing, through became problematic due to the previous lesion in this
mnemonic processes and their neural representations, area. This experiment indicates that the transfer of WM
develop at the same time during childhood and adoles- processes between different brain areas is programmed
cence, and into young adulthood. Similarly, Klingberg and must be executed during the development. The same
et al. (2002) and Schweinsburg et al. (2005) observed that argument about different functional brain areas of spe-
the increase in visual-spatial WM in the 9–18 age period cialization can be derived from the work by Crone et  al.
is correlated with increased activity in frontal and pari- (2006) on an object WM task. They found that the ventro-
etal areas, as well as progressive maturation, due to the lateral prefrontal cortex participated in the maintenance
myelination of the pathways that interconnect these two of items in the WM in both children and adults, whereas
structures. the dorsolateral prefrontal cortex was involved in the
Focusing only on the development of non-spatial manipulation of items in the WM in children older than
WM in three age groups (8–12, 13–17, and 18–25 years old), 12 and adults. The youngest subjects in the experiment
Crone et al. (2006) observed that the younger group had (8–12  years old) did not recruit the dorsolateral prefron-
lower performance than the other two groups, but no tal cortex for item manipulation. This view of different
differences were observed in the ventrolateral prefron- areas contributing to the WM functions during develop-
tal cortex activation patterns (a region associated with ment has also been observed in a decrease with age in the

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
8      C.M. Gómez et al.: Working memory development

contribution of the hippocampus to WM, whereas the pre- after the presentation of infrequent stimuli (Squires et al.,
frontal cortex becomes more crucial during adolescence 1976; Johnson and Donchin, 1982), and its amplitude is
(Finn et  al., 2010). The maturational trend is consistent inversely proportional to the probability of the stimulus’
with the general view that during development there is occurrence. Donchin and Coles (1988a) proposed that
a more distributed processing that becomes increasingly the posterior P300 or P3b would represent an operation
focused (Durston et al., 2006). of updating the context where the WM operates (Polich,
2007). However, P3b presents alternative interpretations
(Mangun and Hillyard, 1991; Verleger et al., 2005; Polich,

Event-related potentials associated 2007; Gómez and Flores, 2011). Polich (2007) proposed
that the P3b component is related to the neuroinhibition
with WM development that is necessary to focus attention on the relevant task,
facilitating the action of memory systems without dis-
The event related potential (ERP) technique, applied to turbances. In general, a close relationship between the
understand the cerebral mechanisms of WM process- P300 and WM updating is broadly supported. P3a, the
ing of information, is useful for making inferences about other sub-component of P300, is generated as the brain’s
the timing and anatomical location of these processes in response to stimuli that are novel compared to the more
memory. The development of WM depends on several pro- frequent stimulation (Friedman et al., 2001). This compo-
cesses, including the processing of information specific nent would index an attentional reorientation generated
to each modality, its coding and retention in memory, by an unexpected stimulus (Polich, 2007) or a change in
the ability to recognize the stored items, and the updat- the processing rules (Barceló et al., 2002). Likewise, it has
ing of the contents in the WM. These processes can be been proposed that the disconfirmation of the subjective
studied with MEG and EEG, given their excellent temporal expectancy about the characteristics of the target stimulus
resolution. is related to the generation of the P3a component (Gómez
et al., 2008). Frontal lobe lesions decrease the amplitude
of the P3a, but not the P3b recorded in parietal areas

Developmental neurophysiology (Lovstad et al., 2012), indicating a posterior localization of


P3b neural sources. The reduction in P3a after injury is not
of the updating process only caused by damage in the lateral prefrontal cortex, but
also by lesions in the orbitofrontal cortex (Lovstad et al.,
There are quite a few studies on P300 maturation that can 2012). However, lesions in the temporo-parietal junction
be interpreted as a developmental index of the WM con- reduce the amplitude of both components. In sum, the
tent-updating process. Although most of the results have generators of P3a and P3b are rather distributed, but P3a
been obtained using attentional paradigms, the WM state- has a high dependence of the frontal zones and the tem-
based models suggest that the updating of contents in the poro-parietal junction, whereas P3b does not depend on
WM is a critical aspect in understanding WM operations, frontal areas, but rather on posterior regions (Bledowski
and this updating would have a close correlation with et al., 2004).
attentional processes. Therefore, P300  maturation can One type of paradigm generating a P300 component
be considered an important component in the updating that can be linked to the process of WM updating is the
process. Another type of paradigm in ERPs that could offer N-back, where the subject has to compare the current item
insight on the development of WM sub-processes is the with previously presented items (N-1 back, N-2 back, etc.).
DMTS and related experimental paradigms. In DMTS-type The study of the process of change in the WM contents
tests, components related to the time when the stimulus is (Wild-Wall et al., 2011) suggests that the N-back task can
coded, kept in memory (interval between S1 and S2), and be categorized as a dual-task: (i) a WM task related to the
recovered (matching) can be studied. Particularly, aspects process of updating and (ii) a matching task between the
related to the neural signals that make it possible to main- stored item and the new item (Watter et al., 2001).
tain the stored items in the WM, and the negativities asso- However, there is no systematic approach to defining
ciated with the matching process, are reviewed here. the developmental trajectories of the P300 during devel-
The P300 is one of the most complex waves, and it has opment and their relationship with WM (although in some
been related to different functional properties, including instances the P300 component has been used to assess
the updating of WM content. This ERP corresponds to a WM updating in children; Evans et al., 2011). Therefore, we
late positive component (LPC) that appears around 300 ms concentrate now on describing how the P300 component

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      9

evolves with age in other types of paradigms, basically in This decrease in amplitude and latency has also been
the ‘Oddball paradigm,’ where a more extensive descrip- found by other authors (Thomas and Nelson, 1996; Van
tion has been carried out. This is justified, in the context der Stelt et al., 1998). Rojas-Benjumea et al. (2015), record-
of the present review, based on the hypothesis that P300 ing a population from 6 to 26 years old, found a decrease
is related to the process invoked when there is a need to in latency and peak-to-peak amplitude of the P3b compo-
revise the representations in the WM (Donchin and Coles nent. The decrease in amplitude with age in a posterior
1988a,b). positivity, categorized as an LPC, has also been obtained
Several studies have analyzed the amplitude and in DMTS paradigms during the encoding and recognition
latency of the endogenous components P3a and P3b phases (Figure 2B) (Barriga-Paulino et al., 2014, 2017).
during child development in the auditory modality. Using In conclusion, and given the presence of P3b in chil-
the three-stimulus Oddball paradigm (standard, deviant, dren of at least 6 years old, it is possible to conclude that
and novel stimuli), Määttä et  al. (2005) reported results the cognitive function indexed by this component (the
from children between 8 and 9  years old and adults updating of the WM content) is already formed at this age.
between 22 and 28  years old. These authors reported a However, as most of the studies presented in the present
P3a of maximum amplitude in frontal areas in children, report indicate, P3b maturation continues during child-
whereas in adults, maximum P3a was found in the central hood and adolescence. The fact that the visual modality
areas. In another study using the Oddball task, Fuchigami does not present a clear P3a component in childhood and
et  al. (1995) found that the latency of both P3a and P3b early adolescence suggests that detection of novelty, a
decreased with age, and that P3a matures earlier than processing step that precedes WM content updating, must
P3b. Oades et al. (1997) replicated the decrease in latency be assumed, at least partly, by different brain areas in
with age for the P3b component. In general, with regard to children and adults.
auditory P3b, the most frequently reported general trend More detailed experiments based on structural models
is a reduction in latency and an increase in amplitude of WM, taking into account the evolutionary perspective,
from childhood to adulthood (Segalowitz et al., 2010). are needed to understand how the different phases of
A few studies have analyzed P3  maturation in the WM updating (retrieval, transformation, and substitution)
visual modality in children. With a three-stimulus Oddball evolve with age (Ecker et  al., 2010; Linares et  al., 2016).
task, Stige et al. (2007) analyzed two age groups (6.8–15.8 The N-back task would be an excellent paradigm when
and 20–88.8  years), and they found that P3a matures combined with EEG or MEG techniques to explore the
earlier than P3b. They showed that the latency of P3a updating sub-components and to trace the developmental
increased and the amplitude decreased with age. The P3b trajectories of these sub-processes. The different number
component showed a nonlinear reduction in amplitude of sub-processes involved in N-1 (substitution) and N-2
with age. By contrast, Courchesne (1978) found no visual (substitution and retrieval) tasks would facilitate to iden-
component P3a in children, but they reported the presence tify the ERP indices of each sub-process. Other tasks
of P3b. Our group reported the absence of the P3a compo- such as the mental counter task (changing the numerical
nent in children and preadolescents in the visual modality values of stored counters upon certain cues) would permit
of two different types of visual experimental paradigms; us to explore the sub-process of transformation. However,
(i) odd-ball (Figure 1A) (Rojas-Benjumea et al., 2015) and at this point there is no systematic approach of dissecting
(ii) central-cue Posner paradigm (Figure 1B) (Flores et al., these different updating sub-processes with a develop-
2009). One possible reason for the discrepancy between mental perspective.
the developmental trajectories of auditory and visual P3a
would be a possible contribution of posterior dipoles situ-
ated in the inferior bank of the Sylvian fissure for the audi-
DMTS paradigms for understanding the
tory modality, whereas in the visual modality, P3a sources
would be frontally situated. Given the earlier maturation of developmental neurophysiology of the
posterior cortex areas compared to frontal areas, the pres- encoding, maintenance, and recognition
ence of P3a in children and adolescents in the auditory process
modality would be justified. With regard to the P3b com-
ponent, Thomas and Nelson (1996) and Van der Stelt et al. The DMTS task makes it possible to separate the encod-
(1998) found a decreased P3b latency and amplitude with ing, maintenance, and recognition phases of WM, given
increasing age. Flores et al. (2009), in a central cue Posner the considerable separation in time between the retained
paradigm, showed a decrease in P3b with age (Figure 1B). item (S1) and the recognized item (S2). DMTS and DRT

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
10      C.M. Gómez et al.: Working memory development

A B Valid
Time condition
Delay
(82.1%)

Invalid
700 ms condition
Delay
(17.9%)
700 ms
Cue Target Motor Cue
700 ms 300 300 response 300

700 ms
1370 1860
Age groups
Cz 6–9 Children Young adults
P3a FCz
10–13
14–17
18–21
22–26 P3a
Pz Cz P3a

Invalid
10 µV 10 µV Valid
+ +
P7
POz P3b
– –
15 µV 10 µV P3b
P3b
Pz

Oz
O2

S2 S2

0 400 800 0 500 0 500


Time (ms) Time (ms)

Figure 1: P3a and P3b during development.


(A) Above: an example of a trial on the Oddball task. Presentation of a sequence of cartoons where the bee was the frequent standard stimu-
lus and other cartoons were the infrequent novel target stimuli. The subjects had to respond to the novel stimuli. Below: difference waves
obtained by subtracting the ERPs in the standard condition from the ERPs in the target condition. Notice the presence of P3a from the 14- to
17-year-olds to the oldest group, and the presence of P3b in all age groups, although with a higher latency in the youngest group (adapted
from Rojas-Benjumea et al., 2015). (B) Above: the experimental paradigm comprised two conditions, valid and invalid (82.1% of valid trials).
In the valid condition, the cue indicated the spatial position where the visual target appears. In the invalid condition, the target appears on
the opposite side to the cued location. Below: grand average ERPs to the targets in the valid and invalid conditions for children and young
adults. Note the absence of the P3a component in children, whereas the P3b component was already present in children. Adults presented
both the P3a and P3b components (adapted from Flores et al., 2010).

are paradigms that were first designed and developed modality, these components are similar to those obtained
to work with animals by means of lesions and invasive in adults (Hillyard and Picton, 1987), and they are labeled
recording techniques (Fuster, 2007), facilitating the possi- C1 (the striate component), frontal N1, posterior P1, pos-
ble interpretations of signals collected with non-invasive terior N1 and P2, anterior N2, and parietal P3 (also called
techniques. the LPC for paradigms different from the oddball para-
digms and showing different latencies and/or topogra-
phies from the standard P3). The main difference between
The encoding process children and adults in this encoding phase is a mono-
tonic decrease in the amplitude of all these components
Encoding in DMTS paradigms is observed as the typical (Barriga-Paulino, 2014, 2017). The difference in ampli-
succession of ERPs after S1 presentation (Figure 2A for tude of EEG and ERPs in children is probably not only
the experimental paradigm description). In the visual due to different electromagnetic passive properties of the

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      11

A Time Auditory
S2 feedback
2000 ms

Delay Answer
2000 ms

S1 Matching
1500 ms
Retention
1000 ms
Encoding
S2 Auditory
S1 Delay Answer feedback + S1

1000 ms 1500 ms 2000 ms 600 ms 2000 ms 1000 ms


Encoding Retention Matching Encoding

B Encoding Age groups


Cz 6–9
POz P1 Oz
P2 P1 10–13
12 LPC
0 LPC 14–17
8 20 18–21
22–26
4 C1
µV

–10 10 C1
N1
N2 0
0
–20
0 400 1200 0 400 1200 0 400 1200
S1 Time (ms) S1 S1

Retention
Cz Slow wave P8 POz P2 Slow wave
P1
0 Slow wave
10 5
µV

0 0
–6
–10
1200 2400 1200 2400 1200 2400
Time (ms)

C Encoding Retention
6–9 10–13 14–17 18–21 22–26
2 6 –9 10–13 14–17 18–21 22–26
40–60 ms
–2 10
C1 1100–1130 ms
25 –10
100–130 ms P1 offset
–25
N1 P1
20 10
130–160 ms 1160–1230 ms
–20 –10
N2 P2 offset
15
160–230 ms Slow wave
–15 15
P2 2100–2400 ms –15
25 Slow wave
290–400 ms
–25
LPC

Figure 2: Maturation of ERPs during the encoding and delay (retention) phases in a delayed match-to-sample task.
(A) Example of a trial from the delayed match-to-sample paradigm. (B) ERPs observed in the encoding and retention phases for the five age
groups. Time is expressed from S1 onset. The different ERP components are indicated. Note the polarity inversion of the slow wave in the
youngest group: positive in Cz and negative in P8. (C) Topographical maps of ERPs for the encoding and retention phases in the five age
groups (adapted from Barriga-Paulino et al., 2017).

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
12      C.M. Gómez et al.: Working memory development

skull (Gómez et al., 2017), but also to differential synap- with a duration of 500 ms predominated in the right poste-
tic density (Huttenlocher, 1990; Giedd et  al., 2009). The rior area. In addition, Ruchkin et al. (1997) reported a slow
psychophysiological meaning of these components goes posterior negative wave on a WM task. According to these
beyond the scope of the present review (for a review, Luck studies, the posterior activity observed on tasks involv-
et al., 2000). However, the presence of all the visual ERPs ing visual stimuli and retention of information about the
during the DMTS encoding phase of WM suggests that object suggests that the coding and storage of the stimulus
visual processing in WM tasks is not qualitatively different characteristics activate posterior cortical networks. Based
from other cognitive processes in which visual analysis is on several investigations, Berti et al. (2000) suggest that
needed. For instance, an approach that would be useful frontal activity might reflect motor control or preparation
to indicate which visual ERPs are critical for the encoding processes, whereas slow negative potentials observed in
phase is the retrospective analysis of visual ERPs in which the parieto-occipital area would reflect activity related to
a missing or incorrect response is obtained. This approach the storage of the stimulus. This NSW could be the mac-
has not yet been used during WM development. Another roscopic counterpart of the sustained activity observed
possible approach would be the correlation of the number in single cells recorded in the same types of paradigms
of errors with the amplitude of the visual ERP components (Goldman-Rakic, 1995; Fuster, 2007).
during the development. This approach showed that the Using a DMTS paradigm in a sample of subjects from 6
c1, P1, N1, N2 and LPC components were correlated with to 26 years old, we observed an NSW in the posterior region
errors, and significant correlations were maintained after during the retention period of the memory stimulus. The
controlling for age in the P1, N2, and LPC, indicating the NSW was already present in the youngest subjects. The
importance of early (P1) and late processing (N2 and LPC) NSW was extended to more central areas in older sub-
ERPs for a correct encoding process. jects (Figure 2B and C) (Barriga-Paulino et al., 2014, 2017).
In fact, a slow positive wave in the anterior region was
observed in younger children, becoming extinct with age
The negative slow wave during the retention and resulting in negativity in adults.
period Dipole analysis, using the brain electrical source
analysis program, showed that only two posterior located
Several ERP studies, using S1-S2 paradigms, have revealed dipoles are needed to explain almost all brain activity
a negative slow wave (NSW) during the retention period of during the retention period in children. However, the
the stimulus on WM tasks (Ruchkin et al., 1990, 1992). The adults needed two extra dipoles situated in more ante-
amplitude and brain topography vary depending on the rior locations to explain the scalp topography during the
mode and type of information to be memorized (Barceló retention period. This result shows that children recruit
et  al., 1997). The NSW presents greater amplitude in the posterior brain resources to maintain the memory foot-
left hemisphere during phonological memory operations print in the WM (Jonkman et  al., 2003), and adults also
(Rugg, 1984a,b) and greater amplitude in the right hemi- use anterior resources, indicating that children and adults
sphere during visual memory operations (Barrett et  al., share the same neurophysiological mechanism during the
1988). In fact, the latter is the main argument support- stimulus retention period (slow waves), but they use the
ing the idea that NSW can be associated with mnemonic recruited brain areas with different intensities at different
processes, rather than unspecific processes, as suggested ages. Neuroimaging studies have found the activation of
by other authors (Luria et al., 2016). This component has frontal areas in WM paradigms in children (Kwon et  al.,
also been found to be sensitive to the difficulty of the task 2002; Tsujimoto et al., 2004; Klingberg, 2006). Based on
(Ruchkin et al., 1992), and it appears to present a different this, the possibility that this positive frontal side of the
topographic distribution on spatial WM tasks and object NSW is not due to the positive side of posterior dipoles
memorization (Ruchkin et al., 1997). should be acknowledged; and the positive frontal side
The memorization of visual material is generally asso- of the NSW in children may represent activity in frontal
ciated with a centro-posterior negativity (Patterson et al., areas. In fact, the regression with age of the neural
1991; Ruchkin et  al., 1992). More specifically, memoriza- sources of the ERPs during a DMTS task (including the
tion of one object has been related to great negativity in retention period), as computed with LORETA (Pasqual-
the fronto-medial area (electrodes F3, Fz, F4) during the Marqui, 2002), has shown that there is a displacement of
retention period (Mecklinger and Pfeifer, 1996). However, a the intensity of neural sources from posterior to anterior
study carried out by Löw et al. (1999), using a visual DMTS regions with age (Barriga-Paulino, 2015b). Along these
task, showed that after the presentation of S1, a slow wave lines, not only a different level of activity in anterior and

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      13

posterior areas in children and adults would occur, but attention that arises in a stage of selective attention and
also and in order to explain the different polarity of frontal information processing (Hillyard and Anllo-Vento, 1998).
NSW at different ages, a different organization of cortical The stimuli selection based on non-spatial features (such
layer activation in frontal areas compared to posterior as shape or color) elicits the SN, which begins between 140
areas may be acknowledged in childhood with respect to and 180 ms post-stimulus and persists for another 200 ms
adulthood, explaining the different polarity of sustained or more (Hillyard and Anllo-Vento, 1998). SN is thought to
slow waves at different ages (Bender et al., 2005). In fact, reflect the selection of relevant features at an early level of
other negativities, such as the CNV (Bender et al., 2005) information processing (Wijers et al., 1989). For the SN to
and the Bereitschaftspotential potential (Chiarenza et al., be elicited, the representation of the selected feature must
1983), found a posterior negativity and an anterior posi- be in an active state to permit the matching process. SN
tivity during development, which becomes a negativity in is best observed in difference waves, where the potentials
frontal sites as age progresses. elicited by a stimulus with unattended features are sub-
In the same study (Barriga-Paulino et al., 2014), a close tracted from the potentials elicited by the same stimulus
relationship was found between slow-wave amplitude when these features are attended. Thus, the increase in
and behavioral performance on the three WM components this posterior negativity for stimuli with attended fea-
(central executive, phonological loop and visuo-spatial tures versus unattended features has been associated
sketchpad) measured with the WMTB-C during the devel- with the identification and categorization of the stimulus,
opment, indicating that the functional role of the NSW which involves comparison with representations stored in
may not be exclusively associated with the DMTS test, but memory (Eimer, 1994).
rather with the construct of WM in a broader sense. With regard to the development of the SN, the study
An ERP component that can be related to the NSW is of selective attention to colors, carried out by Van der Stelt
the so-called contralateral delay activity (CDA) (McCol- et al. (1998) with subjects from 7 to 24 years old, showed
lough et  al., 2007). Using a variation of the DMTS para- an SN distributed by the occipital, temporal, and parietal
digm (where subjects visualized stimuli in both visual regions in the latency range of 150–300 ms. This negativ-
hemifields, but only the items in one of the hemifields ity was more visible in subjects from 19 to 24 and 16 to
had to be remembered), a negativity in the contralateral 18 years old, and less visible for subjects from 13 to 15 and
hemifield to the item to be memorized can be recorded 10 to 12 years old.
by subtracting the ipsilateral activity from the contralat- Another negative component related to the process
eral electrodes. This negative wave presents a posterior of selecting one stimulus surrounded by distractors is
topography, with maximum amplitude at posterior sites the N2 posterior contralateral (N2pc) component. The
(Woodman and Vogel, 2008), and an increase in ampli- deployment of attention to an object in a hemifield pro-
tude with the number of stimuli stored, suggesting that it vokes an inter-hemispheric imbalance of activity between
is related to the process of maintaining the stored items the contralateral and ipsilateral posterior visual areas.
active in the WM during the delay period. Given the nega- This cortical imbalance is measured (between 200 and
tive polarity and the posterior distribution of the CDA, it 300  ms post-stimulus onset) by subtracting the ERPs in
can be considered to have a similar functional role to the the ipsilateral hemisphere to the hemifield where the item
NSW during the retention period. Regarding the develop- to be selected appears from the ERPs in the contralateral
mental trajectory of the CDA, it has not yet been traced, hemisphere (Luck and Hillyard, 1994). The N2pc presents
and the component is described in limited samples of a posterior scalp distribution with maximum amplitude
children and adolescents (Sander et al., 2011; Spronk et al. around electrodes PO7 for targets presented in the right
2013). visual field (RVF), and PO8 for targets presented in the left
visual field (LVF). Neural generators of this component
have been obtained in the extrastriate visual cortex, with
Selection negativity during the stimulus parietal contributions (Hopf et al., 2000). The presence of
recognition phase the N2pc component in the matching phase of the change
detection paradigms suggests that the process of noticing
The matching or recognition phase of the DMTS task cor- the difference between the new input and the stored item
responds to the active search for correspondence between is very similar to the selection process indexed by the N2pc
the stored item maintained in memory and the presented component (Hyun et al., 2009). The obtained contralateral
items during the recognition phase. The selection nega- SN and N2pc would be considered the macroscopic coun-
tivity (SN) consists of an endogenous potential related to terparts of the results obtained by Chelazzi et  al. (1993,

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
14      C.M. Gómez et al.: Working memory development

2001), which showed an increase in neuronal activity in to the hemifield where the stimulus to be matched was
the temporal cortex of monkeys coding for the selected presented. A negative potential, with similar latency to
item presented on the contralateral side during the match- the SN, was obtained in the contralateral hemisphere
ing phase. The N2pc related to object selection seems to (Figure  3E–G) when the ERPs were subtracted (as indi-
already be operational in the 9- to 12-year-old group, cated in Figure 3A–D) (Barriga-Paulino et  al., 2015a).
presenting a slightly higher peak latency than in adults This contralateral negativity (Figure 3E–G), analyzed in a
(Couperus et  al., 2015). Meanwhile, during the encod- population between 6 and 26 years old, was obtained in
ing and maintenance phases, there also seems to be an the difference waves of the ERPs (in electrode P7) induced
N2pc component in children (by using cues and retrocues, by the display where the target stimulus appeared in the
respectively) (Shimi et al., 2014, 2015). N2pc latency was right visual hemifield minus the induced ERPs when the
longer in children than in adults, and, interestingly, chil- target stimulus appeared in the left visual hemifield (and
dren who presented N2pc more similar to adults also pre- vice versa for P8). It was characterized as an SN because
sented greater WM capacity. These latter results indicate the protocol to obtain it was more similar to the one used
the importance of the selective mechanisms, indexed by to obtain the N2pc. However, this SN may share some
N2pc, in prioritizing the encoding of certain items in WM. common characteristics with the N2pc.
A DMTS experiment where the target to be remem- This contralateral negativity to the relevant stim-
bered appeared in the center of the screen, and after a ulus showed a tendency to decrease in latency and
delay period, two different stimuli appeared in the left increase in amplitude with age. In fact, the duration of
and right visual hemifields (Figure 3A and B), recorded a this component was longer in children than in the other
lower positive potential in the hemisphere contralateral age groups (Figure  3E–G). These effects represented

A B
Time Auditory Time Auditory
S2 feedback feedback
2000 ms 2000 ms
Delay Answer Delay E F
2000 ms 2000 ms 6 7 8 9 years old
Matching Matching 6 7 years old
S1 S1
1500 ms 1500 ms 10 11 12 13 years old 8 9 years old
Retention Retention 14 15 16 17 years old 10 11 12 13 years old
1000 ms 1000 ms 18 19 20 21 years old
Encoding Encoding
22 23 24 25 26 years old

C D 5 5
P7 P7
ERPs in electrode P7 Difference wave (B-A)
SN + SN
+
0 0
µV
µV

+ + SN – –
µV

µV

– S2 – S2
S2 S2
–5 –5
0 Time (ms) 0 Time (ms) 0 1000 (ms) 0 1000 (ms)

G
6–9 years old 10–13 years old 14–17 years old 18–21 years old 22–26 years old

CSD

0.02 µV/cm2

Figure 3: Maturation of the selection negativity (SN) during the recognition (matching) phase.
(A, B) Example of trials from the delayed match-to-sample paradigm used in the experiment. (C) Schematic representation of the ERPs obtained
in P7 when the relevant stimulus appeared on the left and on the right of the screen. (D) The SN obtained from the difference waves (left-pre-
sented target minus right-presented target). (E) Difference waveforms of ERPs obtained from RVF stimulation minus LVF stimulation for the five
age groups in electrode P7 during the recognition of the target stimulus. (F) Same as for (E), but for the three youngest age groups. Note that the
SN extends over a longer time in the youngest age groups. (G) Current source density maps for SN of the five groups in the 150–250 ms tempo-
ral window after the S2 onset. Note the current sink over the P7 electrode in the five age groups. The anterior sinks are not discussed because
they may have presented contamination from eye dipoles (adapted from Barriga-Paulino et al., 2015a).

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      15

neurophysiological evidence about the greater effec- component extracted from behavioral results, which
tiveness of the visual selection processes in the transi- would suggest a possible causal relationship between
tion from childhood to adolescence (Van der Stelt et  al., ERPs and behavioral maturation in DMTS.
1998). The results suggest that children, adolescents, and
young adults use the same mechanism and brain areas
(contralateral parietal negativity, categorized as an SN)
to identify and select the recognized stimulus; however, Brain oscillations associated with
children spend more time on this process, possibly due to
the typical immaturity of neural structures.
WM development

Oscillations evoked and induced by sensory


Relationship of ERPs and behavioral stimuli during WM tasks
­maturation in DMTS tasks
Brain oscillations are rhythmic EEG signals generated
In addition to the NSW and the SN, a series of ERPs appear by the activity of connected neurons. Oscillations can
in the encoding, retention (delay), and recognition (match- be studied in resting state or linked to the processing of
ing) phases (Figure 2B and C), including the posterior C1, stimuli or motor responses. The role of the theta band in
P1, and LPC, and the anterior frontal N1 and N2. During WM has been clearly demonstrated; Raghavachari et  al.
the delay period, and in addition to the NSW, offset P1 and (2001) showed an increase in theta band activity during
P2 components were recorded, indicating the processing a Sternberg task. Gamma and theta phase coupling has
of the offset of the encoded stimuli. Children presented been proposed as a mechanism for keeping the items in
higher amplitudes than adults in most of these compo- memory during the retention period of the DMTS task
nents (including the posterior NSW) (Barriga-Paulino (Sauseng et  al., 2010), whereas alpha amplitude has
et  al., 2017), and although the decrease in amplitude of been related to the suppression of distracting informa-
these components could be due to passive electromag- tion (Sauseng et  al., 2009). The synchronization of the
netic properties generated by the increase in skull elec- theta phase during the retention period has also been
trical resistivity with age (Hoekema et  al., 2003), it was described. An increase in theta amplitude during the prep-
recently demonstrated that passive electromagnetic skull aration period of a warning task (where the subjects have
bone properties cannot completely explain the reduction to remember the type of trial (no-go, prosaccades, and
in amplitude of the EEG signals with age (Gómez et  al., antisaccades) in order to produce a correct response) has
2017). also been reported (Cordones et al., 2013). The number of
The ERP components that appear during the differ- oscillations during the matching phase increases with the
ent phases of the DMTS task are related to the visual number of items retained in the WM, which suggests that,
processing of the stimuli in different stages, and they are if more attempts are made to recover information, more
correlated with accuracy, indicating that the maturation phase-locked theta cycles are needed (Tesche and Karhu,
of these components co-varies with performance on the 2000). It is not yet clear how specific the theta oscilla-
different WM operations (Barriga-Paulino et  al., 2017). tions are to WM operations, or whether theta corresponds
When principal component analysis is applied to the to a type of activity required from a myriad of cognitive
matrix of the amplitudes of the different ERPs appearing processes. The role of low beta band coherence between
during the DMTS task, the scores on the first principal parietal and prefrontal cortex during DMTS tasks has been
component extracted from the ERPs during develop- observed in intracranial recordings, showing a type of
ment were indexing both the age-related increase in object content-specific relationship (Salazar et al., 2012).
performance accuracy and the individual variability in The latter approach stresses how important higher cogni-
accuracy performance, suggesting that there is an indi- tive functions, as WM, would depend on the inter-areal
vidual maturational factor of ERPs that is related to WM cortical dynamics of different frequency bands (Bressler
performance. The presence of a first principal component and Richter, 2015).
explaining both age and individual variability has also Using a Sternberg task, Harmony (2013) showed a
been obtained from behavioral results on WM tasks delta band increase in adults after presentation of the
(Barriga-Paulino et  al., 2016). Therefore, the principal stimuli to be retained. However, children did not show
component obtained from ERP (explaining age and this increase. This difference suggests that delta would be
individual variability) results is probably related to the related to an increase in resistance to distractors in adults.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
16      C.M. Gómez et al.: Working memory development

On the other hand, children showed a higher theta rhythm The relationship of WM with spontaneous
increase than adults. This occurred in the most difficult brain oscillations during development
condition, and it would be due to the higher cognitive
effort made to keep the presented items in memory. In The spontaneous EEG is a continuous activity produced
adults, an increase in synchronization in the alpha range by synaptic potentials not linked to the occurrence of any
(during the encoding phase on WM auditory tasks) and specific event, but associated with the general cognitive
a corresponding desynchronization during recognition enhancement with age. Harmony et  al. (1992) demon-
have been observed. Similar results have been obtained strated that theta rhythm maturation is needed to improve
in children, not only in alpha but also in the theta band cognitive performance with age. The amplitude of this
(Krause et al., 2008). It is difficult to establish the particu- cerebral rhythm, which is prevalent in young children,
lar contribution of the frequency bands to the different decreases progressively (Carretié, 2001; Segalowitz et al.,
sub-processes in WM operations, partly due to the variety 2010). Through the study of correlations between the
of paradigms used. power spectral density of spontaneous EEG and the scores
The P300 component has been related to the process on the three WM components, based on the Baddeley and
of WM updating (Donchin and Coles, 1988), and to an Hitch model (1974): central executive, phonological loop
increase in delta and theta oscillations (Basar-Eroglu and visual-spatial sketchpad, Rodríguez-Martínez et  al.
et al., 1992; Kolev and Yordanova, 1997; Yordanova et al., (2013) observed a close association between the matura-
2000). The role of the theta rhythm in WM can also be tion of theta and low beta rhythms and WM maturation
supported by the fact that the P300 component is related in subjects from 6 to 26  years old. This study revealed
to the delta and theta frequency range as an increase in that the subjects who showed better performance on the
power and phase locking to the stimulus in these two neuropsychological test presented smaller theta rhythm
frequency bands (Digiacomo et  al., 2008). The relation- amplitudes in the spontaneous EEG. The same phenom-
ship between delta oscillations and the development enon was observed (although with a lower correlation) in
of P300  has been demonstrated by the simultaneous the low beta rhythm. Moreover, Rodríguez-Martínez et al.
decrease in the latency of peak delta oscillations and the (2014) showed that there is co-maturation between the
P300 latency with age, suggesting that both are part of theta and beta rhythms in fronto-parietal areas, suggest-
the same phenomenon (Kolev et  al., 1994). Yordanova ing that the maturation of these two rhythms in fronto-
and Kolev (1998), using the oddball auditory task (typi- parietal areas may be important for the maturation of WM.
cally used to generate the P300 component), and analyz-
ing single trials, showed that theta oscillations amplitude
decayed with age, not only during the baseline period (as
would be expected in studies on spontaneous EEG), but
What neurophysiology suggests
also in post-stimulus-induced theta. The brain oscilla- about WM maturation from
tions in the delta range at baseline (spatially filtered by
phase-locking to the stimulus), added to the increase
childhood to adolescence and
in power induced by the stimulus, was expressed in a young adulthood
model by Barriga-Paulino et  al. (2013) to explain the
developmental trend of P300 amplitude. Little is known Neuroimaging studies are suitable for providing insights
about how the maturation of inter-trial synchrony and about how neuroanatomical and functional brain matura-
event-related spectral power, induced in P300 eliciting tion are related to WM maturation. The neurophysiology
tasks, are related to the process of WM updating and its of WM provides information about the evolution of WM
maturation. operations (mechanisms) during the different develop-
There is no doubt that theta, low beta, and delta mental stages. In this regard, and taking into account the
rhythms are emerging as a landmark for WM operations, previously mentioned studies about WM, it seems that
including in the developmental period. This is probably subjects from early childhood to young adulthood share
due to the role of slow oscillation in nesting higher fre- the same mechanisms as adults, although there are func-
quency oscillations, and providing them with an adequate tional differences.
spatio-temporal framework (Buzsáki, 2006). However, a Based on the DMTS studies, the NSW and the SN
complete description of how the maturation of power and were similar from 6 to 26 years old (Barriga-Paulino et al.,
synchrony at different frequencies is related to WM devel- 2015a, 2017). Regarding the NSW, the main differences
opment does not yet exist. between children and adults were: (i) higher amplitude

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      17

in children, possibly due to greater cortical thickness. functions) that would occur during the pre-adolescent
This decrease in amplitude with age is general for all the and adolescent periods.
components in the DMTS task and for P3b in the visual In addition, the main difference in the SN corresponds
oddball task; (ii) the possibility of explaining the NSW in to a longer duration of this component in children than in
children with only two posterior dipoles (whereas adults adults (Barriga-Paulino, 2015a), which may correspond to
need a couple of anterior dipoles); and (iii) the positive a slower pace for the extraction of information from the
slow wave registered in frontal sites in children, instead matching stimulus in children. This maturational delay
of the widespread NSW recorded in adults. The presence would be due to physiological maturational processes that
of an NSW in children clearly suggests a similar mecha- are taking place in young children. The maturation of the
nism for WM during the retention period in children and posterior cortex, where the neural sources of the SN are
adults. However, different dipole modeling and polarity located, is asynchronous in different posterior areas. As
changes indicate that brain activation is not the same Giedd et al. (2009) indicated, a maturational gradient from
in childhood and adulthood. Neuroimaging studies, low to higher-order association cortices inside lobules is
reviewed above, and LORETA analysis of NSW (Barriga- superposed with the more general anterior-­posterior matu-
Paulino, 2015b), suggest the progressive involvement of ration gradient. This within-lobule gradient would justify
the frontal cortex in WM tasks as age increases. However, the higher SN duration in children than in adults if SN has
at the same time, the different polarity in the frontal its neural sources in these late maturation associative areas.
cortex suggests a different organization of the neural acti-
vation in cortical layers. A simplified schema of how EEG
is generated in cortical layers (Bear et  al., 2001) would
suggest that the positive SW in frontal areas in children
Conclusions
would be dominated by near to the soma excitatory post-
From the review of the neuroimaging studies, it can be
synaptic potentials, whereas in adults the frontal NSW
concluded that children and adults use similar brain areas
would be dominated by superficial excitatory postsynap-
(mainly fronto-parietal) to process stimuli in WM. In the
tic potentials and/or near to the soma inhibitory postsyn-
same way, electrophysiological studies show that both age
aptic potentials.
groups use similar brain mechanisms (including NSW, SN
In the same direction, the work of our group (Flores
and P3b), but they suggest that in WM processing, in chil-
et  al., 2009, 2010) has shown that, although children
dren the posterior areas are more important than the ante-
perform attentional tasks with slightly lower efficacy
rior ones, and in adulthood the anterior areas would have
to adults, they present important differences in ERPs,
greater participation. It is necessary to study more in depth
which indicates that they use different brain resources to
how brain oscillations participate in WM maturation.
perform the same tasks. Specifically, the presence of the
orientation frontal sub-component of the CNV in children
Acknowledgments: This work has been possible thanks to
is not evident, nor is the frontal motor negative compo-
grants from the Ministry of Science and Innovation PSI2013-
nent that appears in the late CNV. Likewise, the formation
47506-R and PSI2016-80059-R (FEDER funds from the EU),
of a visual P3a in the processing of target stimuli in the
and from the Consejería de Innovación, Ciencia y Empresa
presence of invalid cues (Flores et  al., 2010) and novel
of the Junta de Andalucía. We thank to prof. ­Santiago Peleg-
stimuli (Rojas-Benjumea et  al. 2015) was not evident in
rina for the careful review of this manuscript.
children. All these results suggest a protracted matura-
tion of the frontal cortex, which explains the lower perfor-
mance on attentional and WM tasks. However, in children
and adults, the sensory component of CNV (induced by
References
sensory cues) and the P3b modulation (generated by
invalidly signaled target stimuli and novel stimuli) were Baddeley, A. (1992). Working memory. Science 255, 556–559.
clearly formed. Therefore, as is likely to occur in attention, Baddeley, A. (1996). The fractionation of working memory. Proc.
part of the WM function is transferred to frontal areas as Natl. Acad. Sci. USA 93, 13468–13472.
age advances. Thus, it is possible to speculate about the Baddeley, A. (1998). Working memory. Acad. Sci. 321, 167–173.
Baddeley, A. (2012). Working memory: theories, models, and contro-
possibility that the considerable increase in clinical psy-
versies. Annu. Rev. Psychol. 63, 1–29.
chopathology that appears in post-adolescence (Paus Baddeley, A.D. and Hitch, G. (1974). Working Memory. The Psychol-
et  al., 2008) could partly be explained by maladaptive ogy of Learning and Motivation: Adv. Res. Theor. Vol. 8. G. H.
functional reorganizations (including WM and attention Bower, ed. (New York: Academic Press), pp. 47–89.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
18      C.M. Gómez et al.: Working memory development

Barceló, F., Martín-Loeches, M., and Rubia, F.J. (1997). Event-related Buzsáki, G. (2006). Rhythms of the Brain (New York: Oxford
potentials during memorization of spatial locations in the audi- ­University Press).
tory and visual modalities. Electroen. Clin. Neurophysiol. 103, Cabeza, R. and Nyberg, L. (2000). Imaging cognition II: an empirical
257–267. review of 275 PET and fMRI studies. J. Cognitive Neurosci. 12,
Barceló, F., Periañez, J.A., and Knight, R.T. (2002). Think differently: 1–47.
a brain orienting response to task novelty. Neuroreport 13, Carretié, L. (2001). Psicofisiología (Madrid, Spain: Pirámide).
1887–1892. Case, R., Kurland, D.M., and Goldberg, J. (1982). Operational effi-
Barrett, S.E., Rugg, M.D., and Perrett, D.I. (1988). Event-related ciency and the growth of short-term memory span. J. Exp. Child
potentials and the matching of familiar and unfamiliar faces. Psychol. 33, 386–404.
Neuropsychologia 26, 105–117. Chelazzi, L., Miller, E.K., Duncan, J., and Desimone, R. (1993). A neu-
Barriga-Paulino, C.I., Flores, A.B., Rodríguez-Martínez, E.I., Chin- ral basis for visual search in inferior temporal cortex. Nature
chilla, C., and Gómez, C.M. (2013). Multivariate and wavelet 363, 345–347.
techniques of spontaneous electroencephalography and event Chelazzi, L., Miller, E.K., Duncan, J., and Desimone, R. (2001).
related potentials during children maturation – the role of Responses of neurons in macaque area V4 during memory-
phase resetting. J. Biomed. Sci. Eng. 6, 669–682. guided visual search. Cereb. Cortex 11, 761–772.
Barriga-Paulino, C.I., Rodríguez-Martínez, E.I., Rojas-Benjumea, Chelonis, J.J., Daniels-Shaw, J.L., Blake, D.J., and Paule, M.G.
M.A., and Gómez, C M. (2014). Slow wave maturation on a (2000). Developmental aspects of delayed matching-to-
visual working memory task. Brain Cognit. 88, 43–54. sample task performance in children. Neurotoxicol. Teratol. 22,
Barriga-Paulino, C.I., Rodríguez-Martínez, E.I., Rojas-Benjumea, 683–694.
M.A., and Gómez, C.M. (2015a). Electrophysiological evidence Chiarenza, G.A., Papakostopoulos, D., Giordana, F., and Guareschi-
of a delay in the visual selection process in youngest children. Cazzullo, A. (1983). Movement-related brain macropotentials
Front. Hum. Neurosci. 9, 622. during skilled performances. A developmental study. Electroen.
Barriga-Paulino, C.I., Rojas-Benjumea, M.A., Rodríguez-Martínez, Clin. Neuro. 56, 373–383.
E.I., and Gómez, C.M. (2015b). Fronto-temporo-occipital activity Conklin, H.M., Luciana, M., Hooper, C.J., and Yarger, R.S. (2007).
changes with age during a visual working memory develop- Working memory performance in typically developing children
mental study in children, adolescents and adults. Neurosci. and adolescents: behavioral evidence of protracted frontal lobe
Lett. 599, 26–31. development. Dev. Neuropsychol. 31, 103–128.
Barriga-Paulino, C.I., Rodríguez-Martínez, E.I., Rojas-Benjumea, Constantinidis, C. and Klingberg, T. (2016). The neuroscience of
M.A., and Gómez, C.M. (2016). Principal component analysis of working memory capacity and training. Nat. Rev. Neurosci. 17,
working memory variables during child and adolescent devel- 438–449.
opment. Spanish J. Psychol. 19, e62, 1–13. Cordones, I., Gómez, C.M., and Escudero, M. (2013). Cortical dynam-
Barriga-Paulino, C.I., Rodríguez-Martínez, E.I., Arjona, A., Morales, ics during the preparation of antisaccadic and prosaccadic eye
M., and Gómez, C.M. (2017). Developmental trajectories of movements in humans in a gap paradigm. PLoS One 8, e63751.
event related potentials related to working memory. Neuro­ Couperus, J.W. and Quirk, C. (2015). Visual search and the N2pc in
psychologia 95, 215–226. children. Atten. Percept. Psycho. 77, 768–776.
Barrouillet, P., Gavens, N., Vergauwe, E., Gaillard, V., and Camos, Courchesne, E. (1978). Neurophysiological correlates of cognitive
V. (2009). Working memory span development: a time-based development: changes in long latency event-related poten-
resource-sharing model account. Dev. Psychol. 45, 477–490. tials from childhood to adulthood. Electroen. Clin. Neuro. 45,
Basar-Eroglu, C., Basar, E., Demiralp, T., and Schürmann, M. (1992). 468–482.
P300-response: possible psychophysiological correlates in Cowan, N. (1995). Attention and Memory: An Integrated Framework.
delta and theta frequency channels. A review. Int. J. Psycho- (New York: Oxford University Press).
physiol. 13, 161–179. Cowan, N. (2010). Multiple concurrent thoughts: the meaning and
Bear, M.F., Paradiso, M.A., and Connors, B.W. (2001). Neurosci- developmental neuropsychology of working memory. Dev.
ence: Exploring the Brain. 2nd ed (Philadelphia: Williams and Neuropsychol. 35, 447–474.
Wilkins/Lippincott). Crone, E.A., Wendelken, C., Donohue, S., Van Leijenhorst, L.,
Bender, S., Weisbrod, M., Bornfleth, H., Resch, F., and Oelkers-Ax, R. and Bunge, S.A. (2006). Neurocognitive development of the
(2005). How do children prepare to react? Imaging ­maturation ability to manipulate information in working memory. PNAS
of motor preparation and stimulus anticipation by late contin- USA 103, 9315–9320.
gent negative variation. Neuroimage 27, 737–752. D‘Esposito, M. and Postle, B.R. (2015). The cognitive neuroscience of
Berti, S., Geissler, H-G., Lachmann, T., and Mecklinger, A. (2000). working memory. Annu. Rev. Psychol. 66, 115–142.
Event-related brain potentials dissociate visual working De Avila, E. (1974). Children‘s transformations of visual information
memory processes under categorical and identical comparison according to nonverbal syntactical rules. Unpublished doctoral
conditions. Cognit. Brain Res. 9, 147–155. dissertation, York University.
Bledowski, C., Prvulovic, D., Hoechstetter, K., Scherg, M., Wibral, Diaz, S. (1974). Cucui scale: technical manual multilingual assess-
M., Goebel, R., and Linden, D.E.J. (2004). Localizing P300 gen- ment program. Stockton Unified District, Stockton, CA.
erators in visual target and distractor processing: a combined Digiacomo, M.R., Marco-Pallarés, J., Flores, A.B., and Gómez, C.M.
event-related potential and functional magnetic resonance (2008). Wavelet analysis of the EEG during the neurocognitive
imaging study. J. Neurosci. 24, 9353–9360. evaluation of invalidly cued targets. Brain. Res. 1234, 94–103.
Bressler, S.L. and Richter, C.G. (2015). Interareal oscillatory syn- Donchin, E. and Coles, M.G.H. (1988a). Is the P300 component
chronization in top-down neocortical processing. Curr. Opin. a manifestation of context updating? Behav. Brain Sci. 11,
Neurobiol. 31 62–66. 355–374.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      19

Donchin, E. and Coles, M.G.H. (1988b). On the conceptual founda- Giedd, J.N., Lalonde, F.M., Celano, M.J., White, S.L., Wallace, G.L.,
tions of cognitive psychophysiology: a reply to comments. Lee, N.R., and Lenroot, R.K. (2009). Anatomical brain magnetic
Behav. Brain Sci. 11, 408–427. resonance imaging of typically developing children and adoles-
Durston, S., Davidson, M.C., Tottenham, N., Galvan, A., Spicer, J., cents. J. Am. Acad. Child Adolesc. Psych. 48, 465–70.
Fossella, J.A., and Casey, B.J. (2006). A shift from diffuse to Goldman-Rakic, P.S. (1971). Functional development of the prefron-
focal cortical activity with development. Developmental Sci. 9, tal cortex in early life and the problem of neuronal plasticity.
1–20. Exp. Neurol. 32, 366–387.
Ecker, U.K., Lewandowsky, S., Oberauer, K., and Chee, A.E. (2010). Goldman-Rakic, P.S. (1995). Cellular basis of working memory.
The components of working memory updating: an experimen- Neuron 14, 477–485.
tal decomposition and individual differences. J. Exp. Psychol. Gómez, C.M. and Flores, A. (2011). A neurophysiological evaluation
Learn. 36, 170–189. of a cognitive cycle in humans. Neurosci. Biobehav. Rev. 35,
Eimer, M. (1994). Sensory gating as a mechanism for visuospatial 452–461.
orienting: electrophysiological evidence from trial-by-trial Gómez, C.M., Flores, A., Ledesma, A., Digiacomo, M.R., and González-
cuing experiments. Percept. Psychophys. 55, 667–675. Rosa, J. (2007). Fronto-parietal networks activation during the
Eriksson, J., Vogel, E.K., Lansner, A., Bergström, F., and Nyberg, L. contingent negative period. Brain Res. Bull. 73, 40–47.
(2015). Neurocognitive architecture of working memory. Neuron Gómez, C.M., Flores, A., Ledesma, A., Digiacomo, M.R., and
88, 33–46. PMID 26447571. González-Rosa, J. (2008). P3a and P3b components associ-
Evans, J.L., Selinger, C., and Pollak, S.D. (2011). P300 as a measure ated to the neurocognitive evaluation of invalidly cued targets.
of processing capacity in auditory and visual domains in spe- Neurosci. Lett. 430, 181–185.
cific language impairment. Brain Res. 1389, 93–102. Gómez, C.M., Rodríguez-Martínez, E.I., Fernández, A., Maestú, F,
Finn, A.S., Sheridan, M.A., Kam, C.L.H., Hinshaw, S., and D‘Esposito Poza, J., and Gómez, C. (2017). Absolute power spectral density
M. (2010). Longitudinal evidence for functional specialization changes in the magnetoencephalographic activity during the
of the neural circuit supporting working memory in the human transition from childhood to adulthood. Brain Topogr. 30, 87–97.
brain. J. Neurosci. 30, 11062–11067. Harmony, T. (2013). The functional significance of delta oscillations
Flores, A.B., Digiacomo, M.R., Meneres, S., Trigo, E., and in cognitive processing. Front. Integr. Neurosci. 7, 83.
Gómez, C.M. (2009). Development of preparatory activity Harmony, T., Marosi, E., Becker, J., Reyes, A., Rodríguez, M.,
indexed by the contingent negative variation in children. Brain ­Bernal, J., Hinojosa, G., and Fernández, T. (1992). Correlación
Cognition 71, 129–140. entre el análisis de frecuencias del EEG y el rendimiento en
Flores, A.B., Gómez, C.M., and Meneres, S. (2010). Evaluation of pruebas de atención selectiva y memoria en niños. Revista
spatial validity-invalidity effects by the P300 component in latina de Pensamiento y Lenguaje 1, 96–103.
children and young adults. Brain Res. Bull. 81, 525–533. Hillyard, S. and Picton, T.W. (1987). Electrophysiology of Cognition.
Friedman, D., Cycowicz, Y.M., and Gaeta, H. (2001). The novelty P3: Handbook of Physiology Section 1: The Nervous System. Vol.
an event-related brain potential (ERP) sign of the brain‘s evalu- 5, Higher Functions of the Brain. F. Plum, ed. (Bethesda, MD:
ation of novelty. Neurosci. Biobehav. R. 25, 355–373. American Physiological Society), pp. 519–584.
Fuchigami, T., Okubo, O., Ejiri, K., Fujita, Y., Kohira, R., Noguchi, Hillyard, S.A. and Anllo-Vento, L. (1998). Event-related brain poten-
Y., Fuchigami, S., Hiyoshi, K., Nishimura, A., and Haradag, K. tials in the study of visual selective attention. Proc. Natl. Acad.
(1995). Developmental changes in P300 wave elicited during Sci. USA 95, 781–787.
two different experimental conditions. Pediatr. Neurol. 13, Hitch, G.J., Woodin, M.E., and Baker, S. (1989). Visual and phonolog-
25–28. ical components of working memory in children. Mem. Cognit.
Fuster, J.M. (1997). Network Memory. Trends Neurosci. 20, 451–459. 17, 175–185.
Fuster, J.M. (1999). Cortical dynamics of memory. Int. J. Psycho- Hoekema, R., Wieneke, G.H., Leijten, F.S.S., and Van Veelen, C.W.M.
Physiol. 35, 155–164. (2003). Measurement of the conductivity of skull, temporarily
Fuster, J.M. (2007). Cortical memory. Scholarpedia, p. 11609. removed during epilepsy surgery. Brain Topogr. 16, 29–38.
Fuster, J.M. and Alexander, G.E. (1971). Neuron activity related to Hopf, J., Luck, S.J., Girelli, M., Hagner, T., Mangun, G.R., Scheich, H.,
short-term memory. Science 173, 652–654. and Heinze, H.J. (2000). Neural sources of focused attention in
Fuster, J.M. and Bressler, S.L. (2014). Past makes future: role of pFC visual search. Cereb. Cortex 10, 1233–1241.
in prediction. J. Cognitive Neurosci. 27, 639–654. Huttenlocher, P.R. (1990). Morphometric study of human cerebral
Fuster, J.M. and Jervey, J.P. (1981). Inferotemporal neurons dis- cortex development. Neuropsychologia 28, 517–527.
tinguish and retain behaviorally relevant features of visual Hyun, J., Woodman, G.F., Vogel, E.K., Hollingworth, A., and Luck, S.J.
stimuli. ­Science 212, 952–955. (2009). The comparison of visual working memory representa-
Fuster, J.M. and Jervey, J.P. (1982). Neuronal firing in the inferotem- tions with perceptual inputs. J. Exp. Psychol. Hum. Percept
poral cortex of the monkey in a visual memory task. J. Neurosci. Perform 35, 1140–1160.
2, 361–375. Jarvis, H.L. and Gathercole, S.E. (2003). Verbal and non-verbal work-
Gathercole, S.E. (1998). The development of memory. J. Child ing memory and achievements on national curriculum tests at
­Psychol. Psyc. 39, 3–27. 11 and 14 years of age. Educ. Child Psychol. 20, 123–140.
Gathercole, S.E. and Pickering, S. (2000). Assessment of working Johnson, R. and Donchin, E. (1982). Sequential expectancies and
memory in six and seven years old children. J. Educ. Psychol. decision-making in a changing environment: an electrophysi-
92, 377–390. ological approach. Psychophysiology 19, 183–200.
Gathercole, S.E., Pickering, S.J., Ambridge, B., and Wearing H. Johnson, J., Fabian, V., and Pascual-Leone, J. (1989). Quantitative
(2004). The structure of working memory from 4 to 15 years of hardware stages that constrain language development. Hum.
age. Dev. Psychol. 40, 177–190. Dev. 32, 245–271.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
20      C.M. Gómez et al.: Working memory development

Jonkman, L.M., Lansbergen, M., and Stauder, J.E.A. (2003). Devel- Luria, R., Balaban, H., Awh, E., and Vogel E.K. (2016). The con-
opmental differences in behavioral and event-related brain tralateral delay activity as a neural measure of visual working
responses associated with response preparation and inhibition memory. Neurosci Biobehav Rev. 62, 100–108.
in a go/nogo task. Psychophysiology 40, 752–761. Määttä, S., Saavalainen, P., Könönen, M., Pääkkönen, A., Muraja-
Kemps, E., Rammelaere, S.D., and Desmet, T. (2000). The develop- Murro, A., and Partanen, J. (2005). Processing of highly novel
ment of working memory: exploring the complementary of two auditory events in children and adults: an event-related poten-
models. J. Exp. Child Psychol. 77, 89–109. tial study. Neuroreport 16, 1443–1446.
Klingberg, T. (2006). Developmental of a superior frontal-intrapari- Mangun, G.R. and Hillyard, S.A. (1991). Modulations of sensory-evoked
etal network for visuo-spatial working memory. Neuropsycholo- brain potentials indicate changes in perceptual processing during
gia 44, 2171–2177. visual-spatial priming. J. Exp. Psychol. Hum. 17, 1057–1074.
Klingberg, T. (2016). Neural basis of cognitive training and develop- McCollough, A.W., Machizawa, M.G., and Vogel, E.K. (2007). Electro­
ment. Curr. Op. Behav. Sci. 10, 97–101. physiological measures of maintaining representations in
Klingberg, T., Forssberg, H., and Westerberg, H. (2002). Increased visual working memory. Cortex 43, 77–94.
brain activity in frontal and parietal cortex underlies the McElree, B. (2006). Accessing recent events. Psychol. Learn. Motiv.
development of visuospatial working memory capacity during 46:155–200.
childhood. J. Cognitive Neurosci. 14, 1–10. Mecklinger, A. and Pfeifer, E. (1996). Event-related potentials reveal
Kolev, V. and Yordanova, J. (1997). Analysis of phase-locking is topographical and temporal distinct neuronal activation pat-
informative for studying event-related EEG activity. Biol. terns for spatial and object working memory. Cognitive Brain
Cybern. 76, 229–235. Res. 4, 211–224.
Kolev, V., Yordanova, J., and Silyamova, V. (1994). The relation Miles, C., Morgan, M.J., Milne, A.B., and Morris, E.D.M. (1996).
between the endogenous P3 wave and evoked frequency com- Developmental and individual differences in visual memory
ponents in children. J. Psychophysiol. 3, 277. span. Curr. Psychol. 15, 53–67.
Krause, C.M., Boman, P.A., Sillanmäki, L., Varho, T., and Holopainen, Myers, N.E., Stokes, M.G., and Nobre, A.C. (2017). Prioritizing infor-
I.E. (2008). Brain oscillatory EEG event-related desynchroni- mation during working memory: beyond sustained internal
zation (ERD) and synchronization (ERS) responses during an attention. Trends Cogn. Sci. 21, 449–461.
auditory memory task are altered in children with epilepsy. Oades, R.D., Ditteann, B.A., and Zerbin, D. (1997). Development
Seizure 17, 1–10. and topography of auditory event-related potentials (ERP):
Kwon, H., Reiss, A.L., and Menon, V. (2002). Neural basis of mismatch and processing negativity in individuals 8-22 years
protracted developmental changes in visuo-spatial working of age. Psychophysiology 34, 677–693.
memory. Proc. Natl. Acad. Sci. USA 99, 13336–13341. Oberauer, K. (2013). The focus of attention in working memory –
Leavitt, M.L., Diego Mendoza-Halliday, D., and Martinez-Trujillo, J.C. from metaphors to mechanisms. Front. Hum. Neurosci. 7, 673.
(2017). Sustained activity encoding working memories: not fully Ostby, Y., Tamnes, C.K., Fjell, A.M., and Walhovd, K.B. (2011).
distributed. Trends Neurosci. 40, 328–346. Morphometry and connectivity of the fronto-parietal verbal
Lee, S.H. and Baker, C.I. (2016). Multi-voxel decoding and the working memory network in development. Neuropsychologia
topography of maintained information during visual working 49, 3854–3862.
memory. Front. Systems Neurosci. 10, 2. Pascual-Leone, J. and Baillargeon, R. (1994). Developmental
Linares, R., Bajo, M.T., and Pelegrina, S. (2016). Age-related differ- measurement of mental attention. Int. J. Behav. Dev. 17,
ences in working memory updating components. J. Exp. Child 161–200.
Psychol. 147, 39–52. Pascual-Marqui R.D. (2002). Standardized low-resolution brain elec-
Logie, R.H. (1995). Visuo-Spatial Working Memory (Hove: Lawrence tromagnetic tomography (sLORETA): technical details. Methods
Erlbaum Associates), p. 176. Find Exp Clin Pharmacol. 24, 5–12.
Logie, R.H., Zucco, G.M., and Baddeley, A.D. (1990). Interference Patterson, J.V., Pratt, H., and Starr, A. (1991). Event-related potential
with visual short-term memory. Acta Psychol. 75, 55–74. correlates of the serial position effect in short-term memory.
Lovstad, M., Funderud, I., Lindgren, M., Endestad, T., Due-Tønnessen, Electroen. Clin. Neurophysiol. 78, 424–437.
P., Meling, T., Voytek, B., Knight, R.T., and Solbakk, A.K. (2012). Paule, M.G., Bushnell, P.J., Maurissen, J.P.J., Wenger, G.R., ­Buccafusco,
Contribution of subregions of human frontal cortex to novelty J.J., Chelonis, J.J., and Elliott, R. (1998). The use of delayed
processing. J. Cognit. Neurosci. 24, 378–395. matching-to-sample procedures in studies of short-term memory
Löw, A., Rockstroh, B., Cohen, R., Hauk, O., Berg, P., and Maier, W. in animals and humans. Neurotoxicol. Teratol. 20, 493–502.
(1999). Determining working memory from ERP topography. Paus, T., Keshavan, M., and Giedd, J.N. (2008). Why do many psychi-
Brain Topogr. 12, 39–47. atric disorders emerge during adolescence? Nat. Rev. Neurosci.
Luciana, M., Conklin, H.M., Hooper, C.J., Yarger, R.S. (2005). The 9, 947–57.
development of nonverbal working memory and executive con- Pelegrina, S., Lechuga, M.T., García-Madruga, J.A., Elosúa, M.R.,
trol processes in adolescents. Child Dev. 76, 697–712. Macizo, P., Carreiras, M., Fuentes, L.J., and Bajo, M.T. (2015).
Luck, S.J. and Hillyard, S.A. (1994). Spatial filtering during visual Normative data on the n-back task for children and young
search: evidence from human electrophysiology. J. Exp. ­adolescents. Front. Psychol. 6, 1544.
­Psychol. Hum. 20, 1000–1014. Perlman, S.B., Huppert, T.J., and Luna, B. (2016). Functional near-
Luck, S.J. and Vogel, E.K. (2013). Visual working memory capacity: infrared spectroscopy evidence for development of prefrontal
from psychophysics and neurobiology to individual differences. engagement in working memory in early through middle child-
Trends Cogn. Sci. 17, 391–400. hood. Cereb. Cortex 26, 2790–2799.
Luck, S.J., Woodman, G.F., and Vogel, E.K. (2000). Event-related Pickering, S.J. (2001). Cognitive approaches to the fractionation of
potential studies of attention. Trends Cogn. Sci. 4, 432–440. visuo-spatial working memory. Cortex 37, 457–473.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
C.M. Gómez et al.: Working memory development      21

Pickering, S.J. and Gathercole, S.E. (2001). Working Memory Test Schweinsburg, A.D., Nagel, B.J., and Tapert, S.F. (2005). fMRI reveals
Battery for Children (WMTB-C) (London: Pearson). alternation of spatial working memory networks across adoles-
Pickering, S.J., Gathercole, S.E., and Peaker, S.M. (1998). Verbal and cence. J. Int. Neuropsych. Soc. 11, 631–644.
visuo-spatial short-term memory in children: evidence for com- Segalowitz, S.J., Santesso, D.L., and Jetha, M.K. (2010). Electrophys-
mon and distinct mechanisms. Mem. Cognit. 26, 1117–1130. iological changes during adolescence: a review. Brain Cognit.
Polich, J. (2007). Updating P300: an integrative theory of P3a and 72, 86–100.
P3b. Clin. Neurophysiol. 118, 2128–2148. Simmering, V.R. and Perone, S. (2013). Working memory capacity as
Raghavachari, S., Kahana, M.J., Rizzuto, D.S., Caplan, J.B., Kirschen, a dynamic process. Front. Psychol. 3, 567.
M.P., Bourgeois, B., Madsen, J.R., and Lisman, J.E. (2001). Gat- Shimi, A., Kuo, B.C., Astle, D.E., Nobre, A.C., and Scerif, G. (2014).
ing of human theta oscillations by a working memory task. J. Age group and individual differences in attentional orienting
Neurosci. 21, 3175–3183. dissociate neural mechanisms of encoding and maintenance in
Reynolds, G.D. and Romano, A.C. (2016). The development of visual STM. J. Cognitive Neurosci. 26, 864–877.
attention systems and working memory in infancy. Front. Syst. Shimi, A., Nobre, A.C., and Scerif, G. (2015). ERP markers of target
Neurosci. 10, 15. selection discriminate children with high vs. low working
Rodríguez-Martínez, E.I., Barriga-Paulino, C.I., Rojas-Benjumea, memory capacity. Front. Systems Neurosci. 9, 153.
M.A., and Gómez, C.M. (2013). Spontaneous theta rhythm and Spronk, M., Vogel, E.K., and Jonkman, L.M. (2013). No behavioral or
working memory co-variation during child development. Neuro- ERP evidence for a developmental lag in visual working memory
sci. Lett. 550, 134–138. capacity or filtering in adolescents and adults with ADHD. PLoS
Rodríguez-Martínez, E.I., Barriga-Paulino, C.I., Rojas-Benjumea, M.A., One 8, e62673.
and Gómez, C.M. (2014). Co-maturation of theta and low-beta Squires, K.C., Wickens, C., Squires, N.K., and Donchin, E. (1976).
rhythms during child development. Brain Topogr. 28, 250–260. Effect of stimulus sequence on waveform of cortical event-
Rojas-Benjumea, M.A., Barriga-Paulino, C.I., Rodriguez-Martinez, related potential. Science 193, 1142–1146.
E.I., and Gomez, C.M. (2015). Development of behavioral Stewart, L. and Pascual-Leone, J. (1992). Mental capacity constraints
parameters and ERPs in a novel-target visual detection para- and the development of moral reasoning. J. Exp. Child Psychol.
digm in children, adolescents and young adults. Behav. Brain 54, 251–287.
Funct. 11, 22. Stige, S., Fjell, A.M., Smith, L., Lindgren, M., and Walhovd, K.B.
Ross-Sheehy, S., Oakes, L.M., and Luck, S.J. (2010). Exogenous (2007). The development of visual P3a and P3b. Dev. Neuro­
attention influences visual short-term memory in infants. Dev. psychol. 32, 563–584.
Sci. 14, 490–501. Tamnes, C.K., Walhovd, K.B., Grydeland, H., Holland, D., Østby,
Ruchkin, D.S., Johnson Jr. R., Canoune, H., and Ritter, W. (1990). Y., Dale, A.M., and Fjell, A.M. (2013). Longitudinal working
Short-term memory storage and retention: an event-related brain memory development is related to structural maturation
potential study. Electroen. Clin. Neurophysiol. 76, 419–439. of frontal and parietal cortices. J. Cognitive Neurosci. 25,
Ruchkin, D.S., Johnson Jr. R., Grafman, J., Canoune, H., and Ritter, 1611–1623.
W. (1992). Distinctions and similarities among working memory Tesche, C.D. and Karhu, J. (2000). Theta oscillations index human
processes: an event-related potential study. Cognitive Brain hippocampal activation during a working memory task. Proc.
Res. 1, 53–66. Natl. Acad. Sci. USA 97, 919–924.
Ruchkin, D.S., Johnson Jr. R., Grafman, J., Canoune, H., and Ritter, Thomas, K.M. and Nelson, C.A. (1996). Age-related changes in the
W. (1997). Multiple visuospatial working memory buffers: evi- electrophysiological response to visual stimulus novelty: a
dence from spatiotemporal patterns of brain activity. Neuropsy- topographical approach. Electroen. Clin. Neurophysiol. 98,
chologia 35, 195–209. 294–308.
Rugg, M.D. (1984a). Event-related potentials and the phonological Thomas, K.M., King, S.W., Franzen, P.L., Welsh, T.F., Berkowitz, A.L.,
processing of words and non-words. Neuropsychologia 22, Noll, D.C., Birmaher, V., and Casey, B.J. (1999). A developmen-
435–443. tal functional MRI study of spatial working memory. Neuro­
Rugg, M.D. (1984b). Event-related potentials in phonological match- Image 10, 327–338.
ing tasks. Brain Lang. 23, 225–240. Todor, J.I. (1979). Developmental differences in motor task
Salazar, R.F., Dotson, N.M., Bressler, S.L., and Gray, C.M. (2012). ­performance integration: a test of Pascual-Leone’s theory
Content specific fronto-parietal synchronization during visual of constructive operators. J. Exp. Child Psychol. 28, 314–322.
working memory. Science 338, 1097–1100. Tsujimoto, S., Yamamoto, T., Kawaguchi, H., Koizumi, H., and
Sander, M.C., Werkle-Bergner, M., and Lindenberger, U. (2011). ­Sawaguchi, T. (2004). Prefrontal cortical activation associ-
Contralateral delay activity reveals life-span age differences ated with working memory in adults and preschool children:
in top-down modulation of working memory contents. Cereb. an event-related optical topography study. Cereb. Cortex 14,
Cortex 21, 2809–2819. 703–712.
Sauseng, P., Klimesch, W., Heise, K., Gruber, W., Holz, E.M., Karim, Van der Stelt, O., Kok, A., Smulders, F.T.Y., Snel, J., and
A.A., Glennon, M., Gerloff, C., Birbaumer, N., and Hummel, F.C. Gunning, W.B. (1998). Cerebral event-related potentials
(2009). Brain oscillatory substrates of human visual short-term associated with selective attention to color: developmental
memory capacity. Curr. Biol. 19, 1846–1852. changes from childhood to adulthood. Psychophysiology 35,
Sauseng, P., Griesmayr, B., Freunberger, R., and Klimesch, W. 227–239.
(2010). Control mechanisms in working memory: a possible Verleger, R., Jaśkowski, P., and Wascher, E. (2005). Evidence of an
function of EEG theta oscillations. Neurosci. Biobehav. Rev. 34, integrative role of P3b in linking reaction to perception.
1015–1022. J. Psychophysiol. 19, 150.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM
22      C.M. Gómez et al.: Working memory development

Vestergaard, M., Madsen, K.S., Baaré, W.F., Skimminge, A., Ejersbo, Wilson, J.T.L., Scott, J.H., and Power, K.G. (1987). Developmental dif-
L.R., Ramsøy, T.Z., Gerlach, C., Akeson, P., Paulson, O.B., and ferences in the span of visual memory for pattern. Brit. J. Dev.
Jernigan, T.L. (2011). White matter microstructure in supe- Psychol. 5, 249–255.
rior longitudinal fasciculus associated with spatial working Woodman, G.F. and Vogel, E.K. (2008). Selective storage and
memory performance in children. J. Cognitive. Neurosci. 23, maintenance of an object’s features in visual working memory.
2135–2146. Psychon. Bull. Rev. 15, 223–229.
Wild-Wall, N., Falkenstein, M., and Gajewski, P.D. (2011). Age-related Yordanova, J. and Kolev, V. (1998). Developmental changes in the
differences in working memory performance in a 2-back task. theta response system: a single sweep analysis. J. Psycho-
Front. Psychol. 2, 186. physiol. 12, 113–126.
Watter, S., Geffen, G.M., and Geffen, L.B. (2001). The n-back as a Yordanova, J., Devrim, M., Kolev, V., Ademoglu, A., and Demiralp,
dual-task: P300 morphology under divided attention. Psycho- T. (2000). Multiple time-frequency components account for
physiology 38, 998–1003. the complex functional reactivity of P300. Neuroreport 11,
Wijers, A.A., Lamain, W., Slopsema, J.S., Mulder, G., and 1097–1103.
Mulder, L.J.M. (1989). An electrophysiological investigation Zhou, X., Zhu, D., Katsuki, F., Qi, X.L., Lees, C.J., Bennett, A.J.,
of the spatial distribution of attention to colored stimuli in Salinas, E., Stanford, T.R., and Constantinidis, C. (2014). Age-
focused and divided attention conditions. Biol. Psychol. 29, dependent changes in prefrontal intrinsic connectivity. Proc.
213–245. Natl. Acad. Sci. USA 111, 3853–3858.

Brought to you by | University of Florida


Authenticated
Download Date | 11/29/17 6:27 PM

You might also like