Cdev 13004 PDF
Cdev 13004 PDF
Cdev 13004 PDF
The title for this Special Section is The Developing Brain: Evidence for Plasticity
during Childhood and Adolescence, edited by Amanda E. Guyer, Koraly Perez-
Edgar, and Eveline A. Crone
This study examined differences in visual attention as a function of label learning from 6 to 9 months of age.
Before and after 3 months of parent-directed storybook training with computer-generated novel objects,
event-related potentials and visual fixations were recorded while infants viewed trained and untrained images
(n = 23). Relative to a pretraining, a no-training control group (n = 11), and to infants trained with category-
level labels (e.g., all labeled “Hitchel”), infants trained with individual-level labels (e.g., “Boris,” “Jamar”) dis-
played increased visual attention and neural differentiation of objects after training.
In the 1st year of life, naming objects and people of faces and objects. However, the mechanisms that
influences both perceptual and conceptual develop- contribute to this link are not well understood. In the
ment, allowing infants to make connections between present investigation, the impact of label learning on
words and items in their visual environment. The visual attention was examined using a multimethod
specificity of words that parents use to label faces or longitudinal experiment.
objects for infants influences visual and neural pro- Recently, top-down processing, defined here as
cessing immediately (Scott, 2011; Scott & Monesson, the use of established conceptual knowledge and
2009, 2010) as well as later in childhood (Hadley, prior experience in directing attention, was proposed
Pickron, & Scott, 2015). This previous work suggests as important for the development of early face and
a fundamental link between the development of lan- object processing (Hadley, Rost, Fava, & Scott, 2014).
guage learning and visual perceptual representations By this account, visual attention is constrained by
previous learning and is directed to task-relevant
Funding for this research was provided to Lisa Scott by a
stimulus locations and features. In contrast, attention
National Science Foundation CAREER Award (BCS-1056805; driven by a bottom-up mechanism is directed based
BCS-1560810). We would like to thank Bryan Nguyen (University solely on the perceptual features of the stimuli
of California Los Angeles, UCLA Baby Lab) and Sam Hutten (SR
Research, UK) for technical support, programming, testing, and
(Baluch & Itti, 2011; Sarter, Givens, & Bruno, 2001).
analysis consultation; Dr. Gwyneth Rost (UMASS Amherst, The present investigation examined the top-down
Department of Communication Disorders) and Dr. David Shein- impact of verbal labeling experience on visual atten-
berg (Brown University, Department of Neuroscience) for con-
tributing to stimulus development and design; The Center for
tion during the 1st year of life.
Research on Families (UMass Amherst) and Hillary Hadley (Psy-
chological and Brain Sciences, UMass Amherst) for statistical
consultation; and M. Andrade, E. Arnold, L. Banach, R. Barry- © 2017 The Authors
Anwar, M. Buyukozer-Dawkins, E. Glater, H. Hadley, K. Child Development © 2017 Society for Research in Child Development, Inc.
Nakayassu, and A. Sabol for testing support and relevant discus- This is an open access article under the terms of the Creative Commons
sion. Attribution-NonCommercial License, which permits use, distribution and
Correspondence concerning this article should be addressed to reproduction in any medium, provided the original work is properly cited
Lisa S. Scott, Department of Psychology, 945 Center Drive, PO and is not used for commercial purposes.
Box 112250, Gainesville, FL 32611-2250. Electronic mail may be 0009-3920/2018/8903-0002
sent to lscott@ufl.edu. DOI: 10.1111/cdev.13004
Labels Impact Visual Attention 699
Past investigations have examined the influence found both for infants who knew the labels prior to
of previous experience on infant visual attention coming to the laboratory as well as infants who
(Baldwin & Markman, 1989; Hurley, Kovack-Lesh, learned an object label during a 10-min training ses-
& Oakes, 2010; Kovack-Lesh, McMurray, & Oakes, sion. Object familiarity alone did not elicit enhanced
2014). For example, Hurley et al. (2010) found that activity suggesting that verbal label learning has top-
6-month-old infants who had pets looked signifi- down effects on visual object processing.
cantly longer toward images of novel cats or dogs In another set of investigations, parents read their
than those infants without pets. However, it is infants books with labeled monkey faces (Scott &
unclear why these attentional differences were pre- Monesson, 2010) or strollers (Scott, 2011) from 6 to
sent. One possibility is that infants with pets have 9 months of age. During training infants heard par-
the experience of learning a distinct name for their ents label different monkeys or strollers with either
pet and this previous learning in turn biases (a) individual-level names (e.g., “Jamar” or “Anice”),
their overall visual attention and strategies. (b) a category-level label (i.e., “monkey” or “strol-
In toddlers, attention to visual features and prop- ler”), or (c) no label at all. Occipital–temporal event-
erties of objects is influenced by verbal labeling (e.g., related potential (ERP) components, thought to
Samuelson & Smith, 1999; Smith, Jones, Landau, Ger- reflect perceptual processing (P1, N290, and the
shkoff-Stowe, & Samuelson, 2002). For example, ver- P400), were examined before and after book training.
bal labels led toddlers to attend to the diagnostic Results showed that hearing individual labels
feature of a category and correctly categorize novel resulted in differential perceptual ERP responses
objects based on that feature (Smith et al., 2002). when infants were presented with new exemplars
Naming distinct objects with the same word (but not within trained categories. Differential perceptual
tone) leads infants to form inclusive categories at responses were not present after learning a generic or
both 6 and 12 months (Fulkerson & Waxman, 2007; category label (“monkey” or “stroller”) or after sim-
Waxman & Braun, 2005). Hearing word–object asso- ple exposure and suggests that the specificity of labels
ciations also influences the number of categories impacts the development of visual representations.
infants form within a given scene (Althaus & Wester- These training experiments provide robust neural
mann, 2016; Havy & Waxman, 2016). evidence of top-down effects of verbal labeling on
Changes in visual fixation strategies have previ- perceptual processing. However, it is unclear
ously been found to be important for learning com- whether, in addition to these perceptual changes,
monalities across objects and within categories visual attention and the underlying neural indices of
(Waxman & Markow, 1995). Recent eye-tracking attention are also impacted by labeling.
data indicate that hearing labels directs infants’ The Nc, or negative central component, is one of
visual fixation strategies when viewing novel objects the most frequently studied infant ERP components
(Althaus & Mareschal, 2014; Althaus & Plunkett, and is thought to index aspects of attention, atten-
2016; Best, Robinson, & Sloutsky, 2010). Interest- tional engagement, and stimulus salience (e.g.,
ingly, the type or specificity of labels also impacts Courchesne, Ganz, & Norcia, 1981; de Haan & Nel-
infants’ visual attention. For example, when 12- son, 1997, 1999; Reynolds, 2015; Reynolds &
month-old infants hear the same label for multiple Richards, 2005; Richards, 2003; Richards, Reynolds,
items they attend more toward common features & Courage, 2010). The Nc is recorded over frontal
(Althaus & Mareschal, 2014; Althaus & Plunkett, and central electrode locations and occurs between
2016). In contrast, hearing unique labels results in approximately 300–800 ms after stimulus onset.
greater visual attention toward individuating fea- Prior studies have used the “oddball” ERP task to
tures of each image (Best et al., 2010). These findings make inferences about infant attentional engagement
suggest that label learning differentially directs and and memory as well as their ability to discriminate
impacts attention, and is a key component of early stimuli. Overall, the Nc is greater in response to
concept formation. infrequently presented items if infants can distin-
Neural measures have also been used to examine guish the frequently from the infrequently presented
the impact of labels on visual processing. Electroen- stimuli (e.g., Reynolds & Richards, 2005; Richards,
cephalography (EEG) recordings measure the electri- 2003). The Nc is also associated with behavioral dis-
cal activity of the brain recorded from scalp plays of sustained attention such as longer overall
electrodes. In one study, Gliga, Volein, and Csibra looking time, visual discrimination, and object recog-
(2010) found increased EEG gamma-band activity in nition (Ackles & Cook, 2007; Peltola, Lepp€ anen,
response to labeled objects in 12-month-old infants M€aki, & Hietanen, 2009; Reynolds, 2015; Reynolds,
(Gliga et al., 2010). Interestingly, this effect was Courage, & Richards, 2010; Richards, 2003; Richards
700 Pickron, Iyer, Fava, and Scott
et al., 2010). The magnitude of the Nc response has families were paid $10 and given a small toy for their
also been reported to vary across individuals, such participation. Fifty-nine infants were recruited for
that 6-month-old infants displaying short-look dura- the study. Forty-three infants were recruited at
tions show larger Nc responses to novel objects 6 months of age for storybook training. The remain-
(Guy, Reynolds, & Zhang, 2013). The Nc has been ing 16 were recruited at 9 months of age to partici-
described as an obligatory attentional response (Csi- pate in a no-training control group.
bra, Kushnerenko, & Grossmann, 2008; de Haan, The final sample included 34 infants (17 boys)
Johnson, & Halit, 2003; Nelson & Monk, 2001; who contributed to both eye-tracking and ERP
Richards, 2003), suggesting that it is primarily driven results. Thirty infants were racially identified as
by bottom-up, exogenous mechanisms. The present White or Caucasian, two were Black or African
investigation aimed to determine the extent to which American and Caucasian, one was Black or African
labels have a top-down effect on the infant Nc. American, and one was Asian and Caucasian. One
Although previous studies have linked verbal infant was ethnically identified as Hispanic or
labeling to perceptual learning (Scott, 2011; Scott & Latino. Infants came from English speaking homes
Monesson, 2009, 2010), it is unclear how labeling at (information about additional languages spoken in
different levels of specificity impacts later attention the home was not collected) and from families with
to untrained stimuli. In the present study, 6-month- an average of 3.67 people living at home with an
old infants were randomly assigned to one of two average income of $60–70,000. Infants were
storybook training conditions. The books each randomly assigned to receive either individual-level
contained two different “species” of computer- (n = 12), category-level (n = 11), or no training
generated novel objects that either had unique, (n = 11). Trained participants were tested at
individual-level labels or the same general category- 6 months (Mage = 184.78 days, SD = 9.83) and
level label. Two different species were used to deter- 9 months (Mage = 281.55 days, SD = 7.48 days) of
mine if including a label comparison impacted learn- age. The no-training control group was only tested at
ing. Infants in the training groups (individual, 9 months of age (Mage = 280.27 days, SD = 11.08).
category) were tested before and after 3 months of Twenty-five infants were excluded from the final
storybook training (at 6 and 9 months), and those in analyses because they did not return for their 9-
the no-training control group were only tested at month visit (n = 6), they refused to wear the EEG net
9 months of age. At pretest and posttest, infants (n = 4), there was a computer error (n = 3), or they
completed an eye-tracking serial presentation task did not contribute at least 12 artifact-free ERP trials
and an ERP oddball task. Here it was predicted that to each condition at either their 6- (n = 3) or 9-month
learning labels for novel objects, from 6 to 9 months (n = 9) visit. This 57% retention rate is typical of
of age, would result in label-dependent attention infant longitudinal ERP investigations (Scott, 2011;
differences. More specifically, it was expected that Scott & Monesson, 2010; Stets, Stahl, & Reid, 2012).
individual-level labeling but not category-level
labeling would increase visual attention to trained
Stimuli
objects and generalize to untrained exemplars
within trained categories. Top-down label-depen- Sixteen digitized color photographs of computer-
dent effects were also predicted for the Nc ERP com- generated artificial objects were used as stimuli.
ponent. Two “species” of objects were created using Modo
(Version 601). Two species were used to create a
category comparison within each of the training
books. Across both species, object exemplars varied
Method
in size, shape, orientation, and color pattern. Indi-
The university’s institutional review board viduating features were primarily located in the
approved all methods and procedures used in this lower half of the image on the external appendages.
study. Data collection for this study was completed Eight images (four from species “A” and four from
between September 2012 and September 2013. species “B”) were used for training and were pre-
sented within a storybook (see Figure 1). The other
eight images (four from each species) were not
Participants
included in the storybooks and were used to test
Parents of all infants gave informed consent before generalization of learning at both the pre- and post-
testing. All infants were born full term and had no training visits (see Figures 1 and 2 for stimulus
visual or neurological abnormalities. At each session, examples and Figure S1 for all the stimuli).
Labels Impact Visual Attention 701
Figure 1. Examples of novel objects within the two object “species” and the stories used for the storybooks. Participants were either
sent home with individual-level storybooks (examples on the left) or category-level storybooks (examples on the right). All images were
used for both conditions. [Color figure can be viewed at wileyonlinelibrary.com]
Figure 2. Example trials for the serial image eye-tracking (A) and event-related potential (ERP) oddball (B) tasks. [Color figure can be
viewed at wileyonlinelibrary.com]
pixels wide and viewed at a vertical visual angle of for infrequent objects. At posttest (9 months),
5.02° and a horizontal visual angle of 6.09°. infants in the category-level training group had an
Infants viewed three blocks of 45 trials, with average of 26.73 (SD = 10.95) trials for frequent
object species separated by block. Each block and 16.00 (SD = 3.52) trials for infrequent objects.
included a single, frequently presented untrained Following individual-level training, infants con-
and six infrequently presented untrained object tributed a mean of 22.75 (SD = 7.81) trials for fre-
exemplars (randomly presented). Within the three quent and 13.50 (SD = 1.98) trials for infrequent
blocks, each infant saw two blocks of one species objects. Infants in the control group (i.e., no train-
and one block of the other species. The species seen ing) had an average of 20.00 (SD = 6.05) trials for
twice was counterbalanced across participants and frequently and 15.36 (SD = 2.96) trials infrequently
based on initial analyses suggesting no differences presented objects.
across species, responses to both species (all three
blocks) were averaged within conditions. The fre-
Data Analyses
quently presented object was seen 80% of the time
(36 trials), and the infrequent objects were seen 20% A boxplot with 1.5 interquartile range (IQR) was
of the time (9 trials). Infants completed an average used to determine whether there were any outliers
of 124.35 (SD = 39.56) trials at 6 months and 123.22 within the current sample for both the ERP and
(SD = 41.37) at 9 months. Each image was pre- eye-tracking data. None of the participants were
sented for 500 ms followed by a randomly varying flagged above the threshold of 1.5 IQR (Tukey’s
intertrial interval (ITI) of 1,000–1,200 ms. During Hinges third quartile at 6 months (2,799.5 ms) or
the ITI, infants viewed a gray screen with a white 9 months (4,005.00 ms).
fixation cross in the center (see Figure 2B). For both eye-tracking and ERP analyses, signifi-
ERPs were collected with a 128-channel Geodesic cant effects were followed up with paired-sample t
Sensor Net connected to a DC-coupled 128-channel, tests, which were corrected for multiple compar-
high-input impedance amplifier (Net Amps 300 isons using the Bonferroni method. Both uncor-
TM; Electrical Geodesics, Eugene, OR). Amplified rected and corrected p values are included.
analog voltages (1–100 Hz bandpass) were collected
continuously and digitized at 500 Hz. Individual
Eye-Tracking Serial Presentation Analyses
electrodes were adjusted until impedances were
< 50 kΩ. Postrecording processing was completed Two separate analyses were conducted to assess
with Netstation 4.3 (Electrical Geodesics). EEG was changes in visual attention across groups as well as
first digitally low-pass filtered at 40 Hz, then seg- before to after training. The first was designed to
mented and baseline corrected with a 100-ms pres- examine change over time, and the second to com-
timulus recording interval. Trials were discarded pare the three groups at 9 months. To examine
from analyses if there were more than 12 bad elec- training-related change, total dwell time was sub-
trodes (changing more than 300 lV in the entire mitted to a three-factor 2 9 2 9 2 mixed measures
segment). Individual channels that were consis- multivariate analysis of variance (MANOVA).
tently bad (off-scale on more than 70% of the trials) Within-subjects factors included infant age (6 and
were replaced using a spherical interpolation algo- 9 months) and object condition (trained, untrained).
rithm (Srinivasan et al., 1996). Following artifact Storybook training level (individual, category) was
detection, each trial was visually inspected for noise the between-subjects factor. Second, 9-month-old
and rejected if a significant amount of noise or drift total dwell time was submitted to a two-factor
was present. An average reference was used to mixed measures MANOVA with the between-sub-
minimize the effects of reference site activity and jects factor of group (individual, category, control)
accurately estimate the scalp topography of the and the within-subjects factor of object condition
measured electrical fields. (trained, untrained). Means and standard deviations
Participants with fewer than 12 artifact-free trials are presented in Table S1.
per condition were excluded from analyses. At
6 months, infants who subsequently received cate-
ERP Oddball Analyses
gory-level training had an average of 25.09
(SD = 3.21) trials for frequent and 16.18 (SD = 1.89) Mean amplitude for the frequently and infre-
trials for infrequent objects. Infants in the individ- quently presented object was measured for the Nc
ual-level training group had an average of 28.00 component. The analysis time window and elec-
(SD = 8.11) for frequent and 16.00 (SD = 2.59) trials trode channels were chosen based on visual
704 Pickron, Iyer, Fava, and Scott
Figure 3. Average dwell time (ms) to trained and untrained objects before and after training and for the 9-month-old control group.
Darker colored boxes are 95% confidence intervals, and the dots represent individual subject means. Dwell time fixation maps are aver-
aged for each age, condition, and stimuli. *p < .05 corrected.
Labels Impact Visual Attention 705
1,187.55), p > .05. All nonsignificant differences are amplitude to the infrequently compared to frequently
reported in Table S4. presented objects for the left region (infrequent:
M = 7.67 lV, SD = 8.09; frequent: M = 0.89 lV,
SD = 4.92), t(11) = 3.63, p = .004, p < .05 corrected,
Dwell Time Differences at 9 Months
middle region (infrequent: M = 6.86 lV, SD = 4.56;
There were no significant main effects or interac- frequent: M = 0.35 lV, SD = 3.62), t(11) = 3.95,
tions between training groups (individual, category, p = .002, p < .05 corrected, and right region (infre-
and control) at 9 months (see Table S5 for the com- quent M = 4.12 lV, SD = 5.28; frequent
plete MANOVA results). M = 1.78 lV, SD = 5.82), t(11) = 2.49, p = .030, p = ns
corrected (see Figure 5). These significant differences
were neither found for 6-month-old infants in the indi-
ERP Oddball Results vidual-level training group nor for the pre- or post-
training visits for infants in the category-level training
Training-Related Nc Amplitude Change
group. Nonsignificant results from 6-month pretest
From the four-factor 2 9 2 9 3 9 2 mixed mea- and 9-month category posttest results are reported in
sures MANOVA there were two significant main Table S8.
effects found. First, a significant main effect of object
frequency, F(1, 21) = 16.60, p < .001, g2 = .483, due
Nc Amplitude Differences at 9 Months
to infants’ mean amplitude being significantly larger
(more negative) for infrequently presented objects To determine whether the attention changes pre-
(M = 3.84 lV, SD = 4.21) compared to frequently sent in 9-month-olds after individual-level learning,
presented objects (M = 1.30 lV, SD = 2.80). Sec- but not category-level learning, were different from a
ond, there was a significant main effect of region, F na€ıve group of 9-month-old infants, a separate three-
(2, 20) = 4.15, p = .031, g2 = .293, such that the mid- factor 2 9 3 9 3 mixed measures MANOVA was
dle region (M = 3.43 lV, SD = 3.70) was signifi- conducted comparing the three 9-month-old groups.
cantly greater than the left region (M = 2.210 lV, A main effect of object frequency, F(1, 31) = 12.25,
SD = 3.803) and marginally greater than the right p = .001, g2 = .28, and region, F(2, 30) = 6.13,
region (M = 2.124 lV, SD = 3.871). A complete p = .006, g2 = .29, were found (see Table S9 for
report of the four-factor MANOVA can be found in complete MANOVA results). Nine-month-old
Table S6. infants had a significantly greater mean amplitude
These main effects are qualified by two significant for infrequently (M = 3.49 lV, SD = 5.23) compared
interactions. First, a significant two-way interaction to frequently presented objects (M = 0.159 lV,
between object frequency and training group (indi- SD = 4.54), and the middle region (M = 2.54 lV,
vidual and category), F(1, 21) = 4.58, p = .044, SD = 4.22) was significantly greater than the right
g2 = .179. Follow-up paired comparisons revealed region (M = 0.769 lV, SD = 5.03).
that infants in the individual-level training group These main effects are qualified by two significant
had a significantly greater mean amplitude for the interactions. The first was a two-way interaction
infrequently presented objects (M = 4.87 lV, between object frequency and training group, F(2,
SD = 4.33) relative to frequently (M = 1.11 lV, SD = 31) = 3.93, p = .030, g2 = .202. This interaction was
3.12) presented objects, t(11) = 4.83, p = .001, p < .05 driven by a significantly larger mean Nc amplitude
corrected. This significant difference was not found for the infrequently (M = 6.22 lV, SD = 5.31) rela-
for the category training condition (infrequent: tive to the frequently (M = 0.776 lV, SD = 4.46) pre-
M = 2.81 lV, SD = 4.06, frequent: M = 1.50 lV, sented object for 9-month-old infants trained at the
SD = 2.38; p > .05; see Table S7 for nonsignificant individual level, t(11) = 3.66, p = .004, p < .05 cor-
results). The second significant interaction was a four- rected. Nine-month-olds in the category group did
way interaction between object frequency, region, age, not show significant differences between frequently
and training group, F(2, 20) = 3.77, p = .041, g2 = .274 (M = 0.46 lV, SD = 4.21) and infrequently pre-
(see Figure 4). Follow-up paired t tests were con- sented objects (M = 1.34 lV, SD = 3.27’), p > .05
ducted between frequently and infrequently presented (see Figure 5). In addition, the 9-month-old control
objects for the left, middle, and right regions at both 6 group also did not show any significant differences
and 9 months, and between individual- and category- (frequently presented: M = 0.79 lV, SD = 4.91;
level training. Paired comparisons revealed that only infrequently presented: M = 2.92 lV, SD = 6.56),
9-month-old infants who received individual-level p > .05. All nonsignificant 9-month category and
storybook training displayed significantly larger mean control training group results are reported in
706 Pickron, Iyer, Fava, and Scott
Frequent Infrequent
9 Months (posttraining)
* *
Individual
Category
Amplitude Difference
Left Hemisphere Middle Region Right Hemisphere (Frequent Infrequent)
Figure 4. Infant negative central responses to frequently presented (blue) and infrequently presented (orange) untrained objects before
and after 3 months of individual-level or category-level training across the left, middle, and right regions (averaged electrodes within
each region) of interest. 95% Confidence intervals are plotted across time for each waveform. The topographic distribution of the ampli-
tude difference between frequently presented and infrequently presented stimuli is plotted to the right of the corresponding waveforms
and reflect an average of the amplitude difference within the time window of 470–720 ms. The electrode groups used for analyses are
shown in the upper right corner. Significant effects are highlighted with gray boxes and *p < .05, after correcting for multiple compar-
isons.
*
Amplitude ( V)
Frequent Infrequent
Time (ms)
Figure 5. Infant negative central (Nc) responses at 9 months to frequently presented (blue) and infrequently presented (orange)
untrained objects after 3 months of individual-level or category-level training and in a 9-month no-training control group. Waveforms
reflect averaged electrodes over the left, middle, and right frontal brain regions. 95% Confidence intervals are plotted across time for
each waveform. *p < .05, after correcting for multiple comparisons.
Table S10. The second significant interaction was a g2 = .165. This interaction mirrors the findings
three-way interaction between object frequency, reported above and shown in Figure 4 (see non-
region, and training group, F(4, 62) = 3.07, p = .023, significant results in Table S11). An additional
Labels Impact Visual Attention 707
analysis was run that collapsed the 9-month-old cat- untrained images and therefore are not likely due
egory trained group with the 9-month-old no-train- to bottom-up features of the stimuli. These label-
ing control group, and no significant mean driven effects offer robust support for a recent
amplitude differences between the frequently and hypothesis that suggests a top-down mechanistic
infrequently presented stimuli were present. framework is critical to understanding the develop-
ment of object and face processing (Hadley et al.,
2014).
One unanswered question is whether the category-
Discussion
level labels benefited infants above and beyond no
The present investigation examined the impact of training. We did not find any significant differences
label learning on object attention using eye-tracking between the 9-month-olds trained with category
and ERP methods. Families of 6-month-old infants labels and the no-training control group. The lack of
were sent home with storybooks containing two differences between these two groups suggests that
“species” of novel objects and were asked to read without individual-level training infants may not
these books to their infants for 3 months. Infants notice differences between the object exemplars. It is
were randomly assigned to receive books with also possible that in the absence of individual-level
objects labeled at either the individual level or at names infants are creating perceptually inclusive
the category level. Overall, within-subject ERP and object representations that led to no differences in
eye-tracking results converge and suggest that attention across conditions for either task. Although
infants increased their attention to novel images this possibility remains speculative, future work
from within the trained category after 3 months of incorporating an additional categorization measure
learning to individuate labeled object exemplars. would allow us to contrast individuation and catego-
Infants trained with individual-level labels dis- rization learning, and examine whether category
played an increase in dwell time, from 6 to 9 months training leads to performance advantages for catego-
of age, when viewing trained object exemplars. This rization tasks. It is also possible that the top-down
increase in dwell time generalized to exemplars of effects found here for individual-level labels will not
objects that are within the trained “species,” but were be present for category-level labels when using a cate-
not in the storybooks. These same infants (trained gorization task. Increased categorization abilities
with individual-level labels) exhibited a larger ERP after category-level label training would be consistent
Nc amplitude response to infrequently relative to fre- with previous studies that find shared labels direct
quently presented untrained exemplars of objects attention to shared features (Althaus & Mareschal,
within the trained “species.” Infants who received 2014; Althaus & Plunkett, 2016) and lead to category
category-level storybooks or no training did not formation (Fulkerson & Waxman, 2007; Waxman &
show changes in dwell time or differential ERP Braun, 2005; Waxman & Markow, 1995). Finally, one
responses. These results indicate that the specificity other important factor to consider is whether stimulus
of labels that infants heard over the course of training properties (bottom-up) interact with labels (top-
led to increased attention when presented with both down) to promote infant attention and learning. For
trained and untrained exemplars of objects within example, does the amount of exemplar variability or
trained categories. the number of diagnostic visual features impact
The present findings extend past research related infants’ ability to use labels to differentially direct
to the effects of prior experience on visual attention attention and learning?
(Hurley et al., 2010; Kovack-Lesh et al., 2014) and Consistent with prior visual perceptual ERP
suggest that the specificity of labels function to dif- studies (Scott, 2011; Scott & Monesson, 2010), here
ferentially direct visual attention during learning. infants’ Nc ERP component was uniquely impacted
Consistent with Smith et al. (2002), our findings by previous individual-level label experience. The
suggest that infants use labels to direct their increases in Nc differentiation after label learning
attention and develop either detailed (after individ- also extends our understanding of the Nc and chal-
ual-level training) or general (after category-level lenges the notion that its response is primarily dri-
training) representations of objects within the visual ven by bottom-up stimulus features and obligatory
world. We propose that labels act as a top-down attention. Our results are consistent with the inter-
facilitator for visual attention and that attentional pretation that the Nc is an index of selective and
changes associated with label learning generalize to sustained attention (Reynolds et al., 2010) and fur-
novel exemplars. The current attention effects are ther suggests that top-down factors, such as label
not present prior to training and generalize to learning, modulate the specificity of the Nc.
708 Pickron, Iyer, Fava, and Scott
Although the present results highlight the Althaus, N., & Plunkett, K. (2016). Categorization in
importance of label learning from 6 to 9 months infancy: Labeling induces a persisting focus on com-
of age, it is currently unknown whether this per- monalities. Developmental Science, 19, 770–780. https://d
iod of development also marks a sensitive period oi.org/10.1111/desc.12358
Althaus, N., & Westermann, G. (2016). Labels construc-
or whether the infants are similarly receptive to
tively shape object categories in 10-month-old infants.
labeling experience across development. One pre-
Journal of Experimental Child Psychology, 151, 5–17.
vious investigation found that individual-level https://doi.org/10.1016/j.jecp.2015.11.013
training, from 6 to 9 months of age, was associ- Balas, B., Westerlund, A., Hung, K., & Nelson, C. A. I.
ated with process-specific learning that directs (2011). Shape, color and the other-race effect in the
attention and broadly influences perceptual and infant brain. Developmental Science, 14, 892–900.
cognitive processes 4 years later (Hadley et al., https://doi.org/10.1111/j.1467-7687.2011.01039.x
2015). This previous result highlights the impor- Baldwin, D. A., & Markman, E. M. (1989). Establishing
tance of 6–9 months but still does not rule out the word-object relations: A first step. Child Development,
possibility that learning during other developmen- 60, 381–398. https://doi.org/10.2307/1130984
tal periods may also have long-lasting benefits on Baluch, F., & Itti, L. (2011). Mechanisms of top-down
attention. Trends in Neurosciences, 34, 210–224. https://
development.
doi.org/10.1016/j.tins.2011.02.003
Depending on the specific analysis, the present
Best, C. A., Robinson, C. W., & Sloutsky, V. M. (2010).
investigation has limited statistical power. Although The effects of labels on visual attention: An eye track-
problematic, our sample size is similar to several ing study. Proceedings of the 32nd Annual Conference of
previously published reports within this field the Cognitive Science Society, 32, 1846–1851. https://doi.
(Balas, Westerlund, Hung, & Nelson, 2011; Gre- org/10.1006/cogp.1995.1016
deb€ack et al., 2015; Guy, Zieber, & Richards, 2016; Courchesne, E., Ganz, L., & Norcia, A. M. (1981). Event-
Hoehl, 2015; Kopp & Lindenberger, 2011; Reid, Stri- related brain potentials to human faces in infants. Child
ano, Kaufman, & Johnson, 2004; Scott, 2011; Scott & Development, 52, 804–811. https://doi.org/10.2307/1129080
Monesson, 2010). As a field, we need to strive to Csibra, G., Kushnerenko, E., & Grossmann, T. (2008).
increase power. However, it is also important to Electrophysiological methods of studying infant cogni-
tive development. In C. A. Nelson & M. Luciana (Eds.),
examine development using longitudinal and multi-
Handbook of developmental cognitive neuroscience (2nd ed.,
method approaches, and to replicate previous find-
pp. 247–262). Cambridge, MA: MIT Press.
ings using a variety of approaches. The results de Haan, M., Johnson, M. H., & Halit, H. (2003). Development
reported here are consistent with several previous of face-sensitive event-related potentials during infancy: A
investigations (Hadley et al., 2015; Scott, 2011; Scott review. International Journal of Psychophysiology, 51, 45–58.
& Monesson, 2009, 2010). https://doi.org/10.1016/S0167-8760(03)00152-1
The present results suggest that infants use labels de Haan, M., & Nelson, C. A. (1997). Recognition of the
to differentially learn about and attend to events in mother’s face by six-month-old infants: A neurobehav-
their world. Here we show that infants who were ioral study. Child Development, 68, 187–210. https://doi.
trained with individual-level labels from 6 to org/10.1111/1467-8624.ep9706130494
9 months of age learned to individuate these objects de Haan, M., & Nelson, C. A. (1999). Brain activity differ-
entiates face and object processing in 6-month-old
and applied this learning to new exemplars within
infants. Developmental Psychology, 35, 1113–1121.
the trained object category. The learning reported
https://doi.org/10.1037/0012-1649.35.4.1113
here suggests that infants exhibit a profound ability Fulkerson, A. L., & Waxman, S. R. (2007). Words (but not
to use labels to build perceptual and conceptual tones) facilitate object categorization: Evidence from 6-
representations to direct attention. and 12-month-olds. Cognition, 105, 218–228. https://
doi.org/10.1016/j.cognition.2006.09.005
Gelman, S. A., Chesnick, R. J., & Waxman, S. R. (2005).
References
Mother–child conversations about pictures and objects:
Ackles, P. K., & Cook, K. G. (2007). Attention or memory? Referring to categories and individuals. Child Develop-
Effects of familiarity and novelty on the Nc component ment, 76, 1129–1143. https://doi.org/10.1111/j.1467-
of event-related brain potentials in six-month-old 8624.2005.00840.x
infants. International Journal of Neuroscience, 117, 837– Gliga, T., Volein, A., & Csibra, G. (2010). Verbal labels
867. https://doi.org/10.1080/00207450600909970 modulate perceptual object processing in 1-year-old
Althaus, N., & Mareschal, D. (2014). Labels direct infants’ children. Journal of Cognitive Neuroscience, 22, 2781–
attention to commonalities during novel category learn- 2789. https://doi.org/10.1162/jocn.2010.21427
ing. PLoS ONE, 9, e99670. https://doi.org/10.1371/jour Gredeb€ ack, G., Kaduk, K., Bakker, M., Gottwald, J.,
nal.pone.0099670 Ekberg, T., Elsner, C., . . . Kenward, B. (2015). The
Labels Impact Visual Attention 709
neuropsychology of infants’ pro-social preferences. Reid, V. M., Striano, T., Kaufman, J., & Johnson, M. H. (2004).
Developmental Cognitive Neuroscience, 12, 106–113. Eye gaze cueing facilitates neural processing of objects in 4-
https://doi.org/10.1016/j.dcn.2015.01.006 month-old infants. NeuroReport, 15, 2553–2555. https://d
Guy, M. W., Reynolds, G. D., & Zhang, D. (2013). Visual oi.org/10.1097/00001756-200411150-00025
attention to global and local stimulus properties in 6- Reynolds, G. D. (2015). Infant visual attention and object
month-old infants: Individual differences and event- recognition. Behavioural Brain Research, 285, 34–43.
related potentials. Child Development, 84, 1392–1406. https://doi.org/10.1016/j.bbr.2015.01.015
https://doi.org/10.1111/cdev.12053 Reynolds, G. D., Courage, M. L., & Richards, J. E. (2010).
Guy, M. W., Zieber, N., & Richards, J. E. (2016). The cor- Infant attention and visual preferences: Converging evi-
tical development of specialized face processing in dence from behavior, event-related potentials, and cor-
infancy. Child Development, 87, 1581–1600. https://doi. tical source localization. Developmental Psychology, 46,
org/10.1111/cdev.12543 886–904. https://doi.org/10.1037/a0019670
Hadley, H., Pickron, C. B., & Scott, L. S. (2015). The last- Reynolds, G. D., & Richards, J. E. (2005). Familiarization,
ing effects of process-specific versus stimulus-specific attention, and recognition memory in infancy: An
learning during infancy. Developmental Science, 18, 842– event-related potential and cortical source localization
852. https://doi.org/10.1111/desc.12259 study. Developmental Psychology, 41, 598–615. https://d
Hadley, H., Rost, G. C., Fava, E., & Scott, L. S. (2014). A oi.org/10.1037/0012-1649.41.4.598
mechanistic approach to cross-domain perceptual nar- Richards, J. E. (2003). Attention affects the recognition of
rowing in first year of life. Brain Sciences, 4, 613–634. briefly presented visual stimuli in infants: An ERP
https://doi.org/10.3390/brainsci4040613 study. Developmental Science, 6, 312–328. https://doi.
Havy, M., & Waxman, S. R. (2016). Naming influences 9- org/10.1111/1467-7687.00287
month-olds’ identification of discrete categories along a Richards, J. E., Reynolds, G. D., & Courage, M. L. (2010).
perceptual continuum. Cognition, 156, 41–51. https://d The neural bases of infant attention. Current Directions
oi.org/10.1016/j.cognition.2016.07.011 in Psychological Science, 19, 41–46. https://doi.org/10.
Hoehl, S. (2015). How do neural responses to eyes con- 1177/0963721409360003
tribute to face-sensitive ERP components in young Samuelson, L. K., & Smith, L. B. (1999). Early noun
infants? A rapid repetition study. Brain and Cognition, vocabularies: Do ontology, category structure and syn-
95, 1–6. https://doi.org/10.1016/j.bandc.2015.01.010 tax correspond? Cognition, 73, 1–33. https://doi.org/10.
Hurley, K. B., Kovack-Lesh, K. A., & Oakes, L. M. (2010). 1016/S0010-0277(99)00034-7
The influence of pets on infants’ processing of cat and Sarter, M., Givens, B., & Bruno, J. P. (2001). The cognitive
dog images. Infant Behavior and Development, 33, 619– neuroscience of sustained attention: Where top-down
628. https://doi.org/10.1016/j.infbeh.2010.07.015 meets bottom-up. Brain Research Reviews, 35, 146–160.
Kopp, F., & Lindenberger, U. (2011). Effects of joint atten- https://doi.org/10.1016/S0165-0173(01)00044-3
tion on long-term memory in 9-month-old infants: An Scott, L. S. (2011). Mechanisms underlying the emergence
event-related potentials study. Developmental Science, of object representations during infancy. Journal of Cog-
14, 660–672. https://doi.org/10.1111/j.1467-7687.2010. nitive Neuroscience, 23, 2935–2944. https://doi.org/10.
01010.x 1162/jocn_a_00019
Kovack-Lesh, K. A., McMurray, B., & Oakes, L. M. Scott, L. S., & Monesson, A. (2009). The origin of biases
(2014). Four-month-old infants’ visual investigation of in face perception. Psychological Science, 20, 676–680.
cats and dogs: Relations with pet experience and atten- https://doi.org/10.1111/j.1467-9280.2009.02348.x
tional strategy. Developmental Psychology, 50, 402–413. Scott, L. S., & Monesson, A. (2010). Experience-dependent
https://doi.org/10.1037/a0033195 neural specialization during infancy. Neuropsychologia,
Nelson, C. A., & Monk, C. S. (2001). The use of event- 48, 1857–1861. https://doi.org/10.1016/j.neuropsycholo
related potentials in the study of cognitive develop- gia.2010.02.008
ment. In C. A. Nelson & M. Luciana (Ed.), Handbook of Smith, L. B., Jones, S. S., Landau, B., Gershkoff-Stowe, L.,
developmental cognitive neuroscience (pp. 125–136). Cam- & Samuelson, L. (2002). Object name learning provides
bridge, MA: MIT Press. on-the-job training for attention. Psychological
Pascalis, O., Scott, L. S., Kelly, D. J., Shannon, R. W., Science, 13, 13–19. https://doi.org/10.1111/1467-9280.
Nicholson, E., Coleman, M., & Nelson, C. A. (2005). 00403
Plasticity of face processing in infancy. Proceedings of Srinivasan, R., Nunez, P. L., Tucker, D. M., Silberstein, R.
the National Academy of Sciences of the United States of B., Cadusch, P. J., & Peter, J. (1996). Spatial sampling
America, 102, 5297–5300. https://doi.org/10.1073/pnas. and filtering of EEG with spline Laplacians to estimate
0406627102 cortical potentials. Brain Topography, 8, 355–366.
Peltola, M. J., Lepp€anen, J. M., M€aki, S., & Hietanen, J. K. https://doi.org/10.1007/bf01186911
(2009). Emergence of enhanced attention to fearful faces Stampe, D. M. (1993). Heuristic filtering and reliable cali-
between 5 and 7 months of age. Social Cognitive and bration methods for video-based pupil-tracking systems.
Affective Neuroscience, 4, 134–142. https://doi.org/10. Behavior Research Methods, Instruments, & Computers, 25,
1093/scan/nsn046 137–142. https://doi.org/10.3758/BF03204486
710 Pickron, Iyer, Fava, and Scott
Stets, M., Stahl, D., & Reid, V. M. (2012). A meta-analysis Table S3. Eye Tracking: Three-Factor Multivari-
investigating factors underlying attrition rates in infant ate Analysis of Variance (MANOVA) for Changes
ERP studies. Developmental Neuropsychology, 37, 226– in Dwell Time Between Training Groups
252. https://doi.org/10.1080/87565641.2012.654867 Table S4. Eye Tracking: Follow-Up Paired t Tests
Vitevitch, M. S., & Luce, P. A. (2004). A web-based inter-
for Age 9 Training Interaction
face to calculate phonotactic probability for words and
Table S5. Eye Tracking: Two-Factor Multivariate
nonwords in English. Behavior Research Methods, Instru-
ments, & Computers, 36, 481–487. https://doi.org/10. Analysis of Variance (MANOVA) for Dwell Time
3758/BF03195594 Differences Between Training Groups at 9 Months
Waxman, S. R., & Braun, I. (2005). Consistent (but not Table S6. Event-Related Potential (ERP): Four-
variable) names as invitations to form object categories: Factor Multivariate Analysis of Variance (MAN-
New evidence from 12-month-old infants. Cognition, 95, OVA) for Training, Region, Age, and Object-Related
https://doi.org/10.1016/j.cognition.2004.09.003 Negative Central (Nc) Amplitude Changes
Waxman, S. R., & Markow, D. B. (1995). Words as invita- Table S7. Event-Related Potential (ERP): Follow-
tions to form categories: Evidence from 12- to 13- Up Paired t Test for Negative Central (Nc) Ampli-
month-old infants. Cognitive Psychology, 29, 257–302. tude Interaction of Object Frequency 9 Training
https://doi.org/10.1006/cogp.1995.1016
Table S8. Event-Related Potential (ERP): Follow-
Up Paired t Tests for Negative Central (Nc) Ampli-
tude Four-Way Interaction: Training 9 Infant
Supporting Information
Age 9 Region 9 Object Frequency
Additional supporting information may be found in Table S9. Event-Related Potential (ERP): 9-
the online version of this article at the publisher’s Month-Old Three-Factor Multivariate Analysis of
website: Data and stimuli for this study are freely Variance (MANOVA) Negative Central (Nc) Ampli-
available at: https://nyu.databrary.org/volume/ tude Differences Between Training Groups, Object
546. Frequency, and Region
Figure S1. Computer-Generated Novel Objects Table S10. Event-Related Potential (ERP): 9-
That Were Included in the Study Month-Old Negative Central (Nc) Amplitude Dif-
Table S1. Eye Tracking: Average Dwell Time at ferences Follow-Up Paired t Tests for Two-Way
6 and 9 Months Between Training Groups Interaction Object Frequency 9 Training
Table S2. Event-Related Potential (ERP): Average Table S11. Event-Related Potential (ERP): 9-
Negative Central (Nc) Amplitude to Frequently and Month-Old Follow-Up Paired t Tests Negative Cen-
Infrequently Presented Objects at 6 and 9 Months tral (Nc) Amplitude Differences Three-Way Interac-
Between Training Groups tion of Object 9 Region 9 Training