J Neurophysiol 114: 3036 –3038, 2015.
First published June 18, 2015; doi:10.1152/jn.00342.2015.
Neuro Forum
Feature-based and object-based attention orientation during short-term
memory maintenance
Yixuan Ku1,2
1
Key Laboratory of Brain Functional Genomics, Ministry of Education, Shanghai Key Laboratory of Brain Functional
Genomics, East China Normal University, Shanghai, China; and 2Department of Neurology, University of California, San
Francisco, California
Submitted 6 April 2015; accepted in final form 12 June 2015
attention orientation; retrospective cue; auditory attention; featurebased attention; object-based attention
ATTENTION AND SHORT-TERM MEMORY
(STM) are interdependent
cognitive functions. Studies in the visual domain have revealed
that top-down attention can bias STM processing at various
stages, including expectation, encoding, maintenance, and retrieval (reviewed in Gazzaley and Nobre 2012). For example,
cueing a visual item to be probed retrospectively (retro-cue)
during STM maintenance period dramatically improved the
performance of the cued item (Landman et al. 2003). Nevertheless, similar mechanisms of attention orientation in the
auditory domain were much less explored. During the encoding period of an auditory scene, when attention was directed to
an auditory object, the change-detection ability for that object
was remarkably enhanced compared with the uncued condition, especially when the scene contained more than four
objects (Eramudugolla et al. 2005). Later, Backer and Alain
(2012) discovered that during the STM maintenance period,
attention could be guided toward one of the maintained soundobject representations through a retro-cue, thereby attenuating
the change-deafness, as well.
In a recent study published in The Journal of Neuroscience,
Backer et al. (2015) advanced their findings, using electroencephalography (EEG) to explore the neural mechanisms underlying the orientation of attention in the auditory domain.
They again used a delayed matching-to-sample task with auditory scenes as stimuli. The auditory scene consisted of four
different sounds simultaneously presented at four different
Address for reprint requests and other correspondence: Y. Ku, 675 Nelson
Rising Lane, Rm. 510, Dept. of Neurology, Univ. of California, San Francisco,
CA 94158 (e-mail: [email protected]).
3036
locations. Each sound was obtained from one semantic category (nonspeech human, animal, music, or man-made object)
and presented at one free-field location (⫺90°, ⫺30°, ⫹30°,
⫹90°). Therefore, two features (semantic and spatial) together
formed one auditory object. During the STM delay period, a
visual retro-cue directed attention to one object (informative)
or divided attention among the four objects (uninformative).
Informative retro-cues were further separated into two kinds:
the informative-spatial retro-cue, leading to a judgment if the
category of the cued location changed (semantic decision), and
the informative-semantic retro-cue, leading to a judgment if the
location of the cued category changed (spatial decision). The
uninformative trials were also divided into two kinds (spatial
and semantic) and grouped with the related informative trials
together to form spatial/semantic blocks. All retro-cues were
valid and enhanced the STM task performance, reminiscent of
those situations in the visual domain. Compared with the
uninformative retro-cues, both informative-spatial and informative-semantic retro-cues shortened the reaction time, yet
only the informative-spatial retro-cue increased the accuracy.
The authors used event-related potentials (ERPs) and neural
oscillations to explain the behavioral benefits. They first implemented spatial principal component analysis (PCA) to reduce the dimensionality of 60 EEG electrodes to three major
representative principal components (PCs), which accounted
for over 60% of variance in the pooled data from any two of the
three conditions (ERPs from informative-spatial, informativesemantic or uninformative trials) across all participants and
electrodes. Interestingly, two of the three PCs displayed objectbased attention effects; i.e., both informative retro-cues induced differential PC activity compared with the uninformative
retro-cues, but no significant difference was observed between
the informative-spatial and informative-semantic retro-cues.
Instead, the third PC exhibited feature-based attention effects;
i.e., the informative-spatial retro-cue induced differential PC
activity compared with the informative-semantic retro-cue.
Furthermore, when trials were split into fast and slow trials on
the basis of reaction time, the feature-specific difference between the informative-semantic and informative-spatial conditions in the third PC showed distinctive patterns in the fast vs.
slow trials, indicating that feature-specific processes were dissociable and attention allocated to one feature in an auditory
object was not required to influence the processing of another
feature in the same object.
Neural oscillations in several frequency bands (alpha: 8 –13
Hz; low-beta: 13.5–18 Hz; mid-beta: 18.5–25 Hz; high-beta:
25.5–30 Hz) revealed attention effects, as well. Specifically,
0022-3077/15 Copyright © 2015 the American Physiological Society
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 16, 2017
Ku Y. Feature-based and object-based attention orientation during
short-term memory maintenance. J Neurophysiol 114: 3036 –3038,
2015. First published June 18, 2015; doi:10.1152/jn.00342.2015.—
Top-down attention biases the short-term memory (STM) processing
at multiple stages. Orienting attention during the maintenance period
of STM by a retrospective cue (retro-cue) strengthens the representation of the cued item and improves the subsequent STM performance. In a recent article, Backer et al. (Backer KC, Binns MA, Alain
C. J Neurosci 35: 1307–1318, 2015) extended these findings from the
visual to the auditory domain and combined electroencephalography
to dissociate neural mechanisms underlying feature-based and objectbased attention orientation. Both event-related potentials and neural
oscillations explained the behavioral benefits of retro-cues and favored the theory that feature-based and object-based attention orientation were independent.
Neuro Forum
ORIENTING ATTENTION DURING SHORT-TERM MEMORY
object maintained in STM, some features (which may come
from multiple sensory modalities) of this object are strengthened (Fig. 1, darker colored patches), whereas others remain
the same (Fig. 1, lighter colored patches). Orienting attention
to one feature does not need to facilitate the representation of
another feature of the same object (Fig. 1, thicker vs. thinner
dashed lines). Moreover, neuroimaging studies in the visual
domain have revealed that the general top-down attention
orientation is located in the frontoparietal network, largely the
intraparietal cortex and superior frontal cortex (Corbetta and
Shulman 2002), whereas the feature-based attention is located
in subregions of the posterior parietal cortex (PPC) (Greenberg
et al. 2010). Given additional evidence on modality-independent control of attention in PPC (Shomstein and Yantis 2006),
it is reasonable to hypothesize that the frontoparietal network
would control the multimodal object-based attention orientation and that the PPC would direct the feature-based attention
orientation, further receiving inputs from different sensory
modalities. Future neuroimaging studies, especially in the
auditory and tactile domains, will be critical to test this neural
hypothesis.
Several additional issues could be further addressed in follow-up studies with paradigms similar to that of Backer et al.
(2015). First, the fate of the uncued items is interesting,
because such a topic is still being debated in the visual domain.
Some suggest that the uncued representations are removed
from the memory buffer (Kuo et al. 2012), whereas others
indicate that they are unaffected (Rerko and Oberauer 2013).
Further findings in the auditory domain will complement those
visual results and help to resolve this issue. Neural oscillations
in the alpha band may serve as one candidate of neural
markers, since alpha activities have been generally attributed to
inhibition of task-irrelevant representations during STM (Jensen et al. 2014). In addition, prestimulus alpha activities
associated with the expectation of stimuli have been shown to
play causal roles in regulating perception (Romei et al. 2010)
and influencing subsequent STM performance (Zanto et al.
Fig. 1. Amodal attention orientation during shortterm memory (STM). Round patches with different
colors indicate features represented in STM (red,
visual; blue, auditory; yellow, tactile), and gray
round patches depict objects maintained in STM.
Orienting attention to a cued object (darker gray)
strengthens the representation of this object compared with another uncued object (lighter gray) in
STM. However, it may strengthen some features
connected to this object (thicker dashed lines)
while other features remain (thinner dashed lines).
The connection can be bi-directional, i.e., when
attention is oriented to one feature; the object
representation connected to this feature will be
strengthened but may not affect other feature representations from the same object. Overall, the
object-based and feature-based attention orientations during STM are independent.
J Neurophysiol • doi:10.1152/jn.00342.2015 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 16, 2017
parietal alpha/low-beta/mid-beta event-related desynchronization (ERD) indicated object-based attention effects, and the
time periods for these effects were partly overlapped with those
for ERP results. Meanwhile, the feature-based attention effects
were observed across all frequency bands after ⬃1,000 ms to
the onset of the retro-cues (ERD: semantic ⬎ spatial). These
ERD effects were more prominent in the fast trials, indicating
that more efficient attention orientation was associated with
better performance, and indeed ERD could further explain
individual differences in the behavioral (a combined performance measure, accuracy/reaction time) benefit. For alpha/
low-beta ERD, the neural-behavioral correlation was observed
after both informative retro-cues. However, for mid-beta/highbeta ERD, such correlation was only shown after the semantic
retro-cue. Taking these findings together, alpha/low-beta and
mid-beta/high-beta oscillations seemed to represent objectbased and feature-based (semantic) attention, respectively.
The results reported by Backer et al. (2015) are critical to
unravel the neural mechanism underlying auditory attention
orientation during STM maintenance, which is in line with
previous findings in the visual domain. Additionally, the authors dissociate two kinds of auditory attention orientation
(object-based vs. feature-based) through neural activities of
both ERPs and ERDs. Coincidentally, in a later issue of The
Journal of Neuroscience, Katus et al. (2015) reported the use of
a retro-cue to spatially orient attention to a tactile object during
STM maintenance. These two studies, combined with those
visual findings, support the theory of object-based attention
across sensory modalities (Shinn-Cunningham 2008) and suggest an amodal attention orientation (see Fig. 1) during STM.
It will be interesting to see whether the distinction between
object-based and feature-based attention orientation in the
auditory domain in Backer et al. still holds in other sensory
domains, and to show how the object-based attention orientation will behave if the object consists of features from different
sensory modalities, i.e., a multi-modal object (Fig. 1). From
Backer et al., it is possible that when attention is cued to an
3037
Neuro Forum
3038
ORIENTING ATTENTION DURING SHORT-TERM MEMORY
GRANTS
Y. Ku was supported by National Institute of Aging Grant 5R01AG030395,
China National Key Fundamental Research (973) Program 2013CB329501,
and Shanghai Committee of Science and Technology Award 15ZR1410600.
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the author.
AUTHOR CONTRIBUTIONS
Y.K. conception and design of research; Y.K. interpreted results of experiments; Y.K. prepared figures; Y.K. drafted manuscript; Y.K. edited and
revised manuscript; Y.K. approved final version of manuscript.
REFERENCES
Backer KC, Alain C. Orienting attention to sound object representations
attenuates change deafness. J Exp Psychol Hum Percept Perform 38:
1554 –1566, 2012.
Backer KC, Binns MA, Alain C. Neural dynamics underlying attentional
orienting to auditory representations in short-term memory. J Neurosci 35:
1307–1318, 2015.
Bastos AM, Vezoli J, Bosman CA, Schoffelen JM, Oostenveld R, Dowdall
JR, De Weerd P, Kennedy H, Fries P. Visual areas exert feedforward and
feedback influences through distinct frequency channels. Neuron 85: 390 –
401, 2015.
Corbetta M, Shulman GL. Control of goal-directed and stimulus-driven
attention in the brain. Nat Rev Neurosci 3: 201–215, 2002.
Eramudugolla R, Irvine DRF, McAnally KI, Martin RL, Mattingley JB.
Directed attention eliminates “change deafness” in complex auditory scenes.
Curr Biol 15: 1108 –1113, 2005.
Gazzaley A, Nobre AC. Top-down modulation: bridging selective attention
and working memory. Trends Cogn Sci 16: 128 –134, 2012.
Greenberg AS, Esterman M, Wilson D, Serences JT, Yantis S. Control of
spatial and feature-based attention in frontoparietal cortex. J Neurosci 30:
14330 –14339, 2010.
Jensen O, Gips B, Bergmann TO, Bonnefond M. Temporal coding organized by coupled alpha and gamma oscillations prioritize visual processing.
Trends Neurosci 37: 357–369, 2014.
Katus T, Müller MM, Eimer M. Sustained maintenance of somatotopic
information in brain regions recruited by tactile working memory. J Neurosci 35: 1390 –1395, 2015.
Kuo BC, Stokes MG, Nobre AC. Attention modulates maintenance of
representations in visual short-term memory. J Cogn Neurosci 24: 51– 60,
2012.
Landman R, Spekreijse H, Lamme VAF. Large capacity storage of integrated objects before change blindness. Vision Res 43: 149 –164, 2003.
Luck SJ, Vogel EK. Visual working memory capacity: from psychophysics
and neurobiology to individual differences. Trends Cogn Sci 17: 391– 400,
2013.
Pasternak T, Greenlee MW. Working memory in primate sensory systems.
Nat Rev Neurosci 6: 97–107, 2005.
Rerko L, Oberauer K. Focused, unfocused, and defocused information in
working memory. J Exp Psychol Learn Mem Cogn 39: 1075–1096, 2013.
Romei V, Gross J, Thut G. On the role of prestimulus alpha rhythms over
occipito-parietal areas in visual input regulation: correlation or causation? J
Neurosci 30: 8692– 8697, 2010.
Shinn-Cunningham BG. Object-based auditory and visual attention. Trends
Cogn Sci 12: 182–186, 2008.
Shomstein S, Yantis S. Parietal cortex mediates voluntary control of spatial
and nonspatial auditory attention. J Neurosci 26: 435– 439, 2006.
Tsal Y, Lavie N. Attending to color and shape: the special role of location in
selective visual processing. Percept Psychophys 44: 15–21, 1988.
van Kerkoerle T, Self MW, Dagnino B, Gariel-Mathis MA, Poort J, van
der Togt C, Roelfsema PR. Alpha and gamma oscillations characterize
feedback and feedforward processing in monkey visual cortex. Proc Natl
Acad Sci USA 111: 14332–14341, 2014.
Zanto TP, Chadick JZ, Gazzaley A. Anticipatory alpha phase influences
visual working memory performance. Neuroimage 85: 794 – 802, 2014.
J Neurophysiol • doi:10.1152/jn.00342.2015 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.2 on June 16, 2017
2014). Note that in Backer et al., the alpha power preceding the
spatial retro-cue seems stronger than that in the semantic
situation, which may suggest different levels of expectation to
the two kinds of retro-cues, and these neural differences might
potentially explain the behavioral differences in the two retrocue conditions.
Second, spatial location has long been recognized as more
“special” for attention than other nonspatial features (color,
shape, contrast, etc., in the visual domain), because attention
operates like a “spotlight” (Tsal and Lavie 1988). Furthermore,
nonspatial visual features per se are not homogeneous, and
memory precisions of different features decay with variant
speeds (Pasternak and Greenlee 2005). These differences
among features may further explain why the informativespatial retro-cue tends to have greater behavioral benefit than
the informative-semantic retro-cue in the study by Backer et al.
Future studies with multiple nonspatial auditory retro-cues can
test the behavioral benefit and the corresponding neural activity
among different auditory features.
Third, attention orientation can be achieved via feedforward
or feedback connections (Corbetta and Shulman 2002). Two
recent neurophysiological studies in monkeys have revealed
that within the visual cortical hierarchy, feedforward processing
communicates through higher frequency oscillations (gamma:
40 –90 Hz) and feedback processing through lower frequency
oscillations (alpha and low-beta) (Bastos et al. 2015; van
Kerkoerle et al. 2014). It is notable that in the work of Backer
et al., higher frequency oscillations (mid-beta/high-beta,
though gamma activities are typically difficult to observe in
EEG signals) represent feature-based attention orientation and
lower frequency oscillations (alpha/low-beta) represent objectbased attention orientation. Given the frequency-specific directions of connection (feedforward vs. feedback) (Bastos et al.
2015; van Kerkoerle et al. 2014), is there any relationship
between the feature-based/object-based attention orientation
and the directions of information transfer? It will be noteworthy to see how the feedforward/feedback processing contributes to the feature-based/object-based attention orientation.
Many more aspects from the study of Backer et al. (2015)
are worth investigating: for example, manipulating STM load
in the task to see how the burden of STM influences attention
orientation, using an auditory retro-cue to avoid extra switching processes between different sensory modalities, or implementing a contralateral retro-cue to induce a contralateral delay
activity that is widely used in visual STM studies (Luck and
Vogel 2013), as well as the tactile STM study mentioned
(Katus et al. 2015). To conclude, the study by Backer et al.
opens up new ideas about the research on interaction between
attention and STM, which is at the core of cognition.