ABSTRACT the fMRI signal could reveal neural selectivity under

ABSTRACT

The conventional fMRI paradigms have their
limitation in spatial resolution in which they depend on the concept of subtracting
the activations between different attributes of stimuli by averaging the entire
neural populations. With the introduction fMRI adaptation paradigms, a gradual
reduction of BOLD signal caused by repeated presentations of stimulus, the
study of selectivity of neuronal populations on a sub-voxel scale seems
possible. Not only do recent studies have provided evidence that the fMRI
signal could reveal neural selectivity under the fMRI adaptation paradigms, but
also enabled the investigation of the invariance of the neural responses from
the neural populations within the imaged voxels which in turn enhance the
functional resolution. Several mechanisms, including neural fatigue, mismatch
in the observed and expected stimuli, sharpening model and facilitation model,
have been proposed to account for the induced reduction in BOLD signal. However,
there are still many unknowns for the underlying neural mechanisms of fMRI
adaptation paradigm and further investigations should be considering on the
examination of the neural basis of the paradigm.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

INTRODUCTION

Functional magnetic resonance imaging (fMRI) has
been a useful tool as the gateway of human brain function in which it measures
the blood oxygenation level-dependent (BOLD) signal related to the change of blood flow due to local neural activity. Although
fMRI offers a noninvasive measurement of brain activity that is secondary to
the electrical activity of neuronal firing with
moderate temporal resolution and good spatial resolution when compared with
other neuroimaging techniques, the spatial resolution is limited to a millimeter
scale due to its fundamental property of an indirect measure of neural activity
(Fukuda, 2016). Spatial resolution of an fMRI study indicates the extent to
which it discriminates between nearby locations; and it is measured by the size
of voxels, a three-dimensional rectangular cuboid, whose dimensions are determined
by the slice thickness, area of a slice and the grid imposed on the slice during
the scanning process (Tsougos, 2017). A voxel typically contains many thousands
to possibly millions of neurons with various properties; therefore, attempting
to establish a direct selectivity of their responses is almost implausible.

With the introduction of a new methodological
approach, fMRI adaptation, the study of selectivity of neuronal populations on
a sub-voxel scale seems possible even in the absence of requiring a smaller voxel
size. The fMRI adaptation is a phenomenon in which repeated and prolonged presentation
of the same visual stimulus causes a consistent and gradual reduction in
activation in a given voxel, and the reduction is selective to the particular
characteristics of the repeated stimulus (Larsson, Solomon, & Kohn, 2016).
The application of fMRI adaptation has been widely used in cognitive
neuroscience in recent years, typically in visual processing such as the
studies of direction of motion (Kar & Krekelberg, 2016), spatial processing
(Zimmermann et al., 2016) and faces (Harris, Young, & Andrews, 2014), due
to several potential advantages from the implication of fMRI adaptation effect.
For instance, the signal-to-noise ratio of an image is relatively low limited
by the spatial resolution of the technique itself and vascular processes including
cardiac activity and respiration unrelated to the neuronal function introduce
unwanted signal or noise to the fMRI data in which fMRI adaptation helps to overcome
such problem by identifying the regions that are selective to certain stimuli. As
the application of fMRI adaptation is most commonly observed in studies related
to visual processing, there have been researchers proposing the relevance
between fMRI adaptation and visual priming in which priming refers to
improvement in performance reflected by shortened reaction time and enhanced
accuracy with repeated presentations of stimuli on a behavioral aspect (Ganel
et al., 2006; Xu, Turk-Browne, & Chun, 2007). Considering the similarity of
the nature between fMRI adaptation and visual priming by repeated and prolonged
presentations of a stimulus, however, it would be counterintuitive that a
reduction in cortical response reflected by the decrease in BOLD signal is
correlated with an improvement in behavioral performance.

In an attempt to understand and discuss further on
fMRI adaptation, the current review addresses particularly on the potential
advantages that could be brought by the implication of fMRI adaptation effect
and then followed by the underlying neuronal mechanisms that have been
proposed.

 

POTENTIAL ADVANTAGES

As mentioned earlier, the conventional fMRI
paradigms have their limitation in spatial resolution where they rely on the
concept of subtraction of activation between different stimulus attributes by
averaging the entire neural populations which might give homogenous responses
towards stimulus changes or respond differentially tuned for different stimulus
types (Larsson, Solomon, & Kohn, 2016). Therefore, it would be difficult
and almost impossible to infer the underlying properties of the imaged neural
populations. With the introduction of the fMRI adaptation paradigm to recent
studies by repeated and prolonged presentations of stimuli, neuronal
populations could be studied beyond the limitation in spatial resolution and
the fMRI adaptation paradigm serves as a tool to study the properties of
neuronal populations on a sub-voxel scale. In general, these paradigms capture
the reduction in neural responses when the stimulus has been presented
repeatedly or for a prolonged time in which an increase in neural response
elicited by the change in particular stimulus attributes confirms the role of
neural populations that are tuned to a specific stimulus attributes being
modified (Larsson & Smith, 2011). The fMRI paradigm has been adopted in a
recent study to understand the feature-salience hierarchy in face processing
where participants were asked to detect changes of faces presented repeatedly
with either slight modification or no modification to examine whether different
face feature contribute differentially to the neural signal in face responsive
regions such as the fusiform face area (Lai et al., 2014). Apart from the studies
concerning visual processing which are commonly being introduced with the
technique, other cognitive processing studies have also included such technique
to study neural representations among the neural populations that are selective
for specific stimulus properties. For instance, a study has been conducted on
voice processing to test whether the neurons responding to human voice and
musical stimuli are distinct or overlapping (Armony et at., 2015). More recent
studies have also demonstrated the neural selectivity using fMRI adaptation in
which certain neural populations are tuned to specific visual features, such as
color and orientation. However, there has been debate about the degree of neural
selectivity demonstrated by the fMRI adaptation in early visual cortex and
previous studies have suggested that the conflicting results could be due to
the untuned neurons in V1 (Henry,2013).

The above studies provide evidence that the fMRI
signal could reveal neural selectivity under the fMRI adaptation paradigms and
indeed, such paradigms also allow the investigation of the invariance of the
neural responses from the neural populations within the imaged voxels by tagging
specific neuronal populations within an area in the brain (Barron, Garvert,
& Behrens, 2016). Limited by the spatial resolution and the enormously
large number of neurons in one voxel, it is unable to study the invariant
properties of cortical neurons by conventional fMRI paradigms. In order to
study the neural invariance to certain attributes, stimulus is presented
repeatedly or with a prolonged time until the neuronal population is adapted
followed by the change of property of the stimulus. It is noted that stimuli
could undergo only one transformation at a time to examine the invariance of
the tagged neuronal populations to a specific attribute. The adapted fMRI
signal indicates the invariance of the tagged neurons to a particular attribute
while an increase in fMRI signal recovering from the adapted state implies that
the tagged neurons are sensitive to the particular property being changed (Grill-Spector
& Malach, 2001). Previous study has employed the fMRI adaptation paradigm
to investigate the image-invariant face recognition across familiar and
unfamiliar faces in face-selective regions by presenting either different
images of the same identity or different image of different identities of
familiar and unfamiliar faces (Weibert, 2016). Another research has also
demonstrated the study of neuronal invariances on color specificity in the V4-complex
under the fMRI adaptation paradigm by varying the color of the prime and the
color of the target (Van Leeuwen, 2014).

The adaptation shown by a gradual reduction in BOLD
signal with repeated presentations across changes between stimuli demonstrates common
neural representation invariant to attributes being varied and the recovery
from adaptation shown by an increase in BOLD signal implies the selectivity of
specific stimulus properties among neuronal representations.

PROPOSED NEURONAL MECHANISMS

From previous discussions, it is noted that the
fMRI adaptation effect could be elicited across brain regions with different
perceptual modalities; thereby, it is expected that there would be various
mechanisms involved in the observed effect. Although there have been several
potential advantages brought by the fMRI adaptation paradigm, the underlying neural
mechanisms are not fully understood. Several mechanisms have been proposed to
account for the reduction in neural responses observed after prolonged presentations
of a stimulus, ranging from simple neural fatigue to complex interaction
frameworks; and the suggested mechanisms will be discussed in the following.

With repeated presentations of a stimulus,
reduction of neural activity can be observed and it could be caused by a
reduction of firing rates of neurons due to neural fatigue. In a typical fMRI
adaptation experiment, prolonged and repeated presentations of stimuli are either
identical or with the targeted attributes being varied that aim to produce an
adaptation effect and recovery responses respectively. Given the similarities
of perceptual adaptation experiments studying the habituation effect caused by
stimuli repetitions, the reduction in BOLD signal under the fMRI adaptation
paradigm could possibly be understood in terms of habituation which can be
described as a decrease in the strength of a response due to repeated
presentations of a stimulus. However, such conventional view of fMRI adaptation
paradigm on the reflection of neural fatigue does not fully interpret and cover
the recovery responses after varied properties have been presented. Another
proposed model has suggested the differences observed in identical and varied
conditions as the mismatch in the observed and expected stimuli; and neuronal
evidence has shown that, to certain extent, expectation in stimuli influences
the fMRI adaptation regardless of the stimulus presentation duration but
attention is critical to effects of expectation (Larsson & Smith, 2012).

In fact, a reduction in BOLD signal after stimuli
repetitions does not necessarily indicate neural fatigue or a decline in
performance. On the contrary, such reduction in neural activity under the fMRI
adaptation paradigm may imply a sharpening and facilitation model. The
sharpening model suggests that the reduction in neural activity could possibly
due to an improved processing by inhibiting non-selective neurons that are
initially activated while the selective neurons in the neuronal populations
remains activated; therefore, the number of neurons involved in the
corresponding processing has decreased in order to offer an efficient and sparse
representation across the cortex (Weiner et al., 2010). In other words, the
model explains the fMRI adaptation paradigm as the suppression of non-selective
neurons that are irrelevant to the properties of the visual stimuli while the
activation state of the selective neurons remain unchanged. However, it is unclear whether fewer
neurons involving would lead to a faster processing which corresponds to an
enhancement in discrimination speed in visual priming. Another
possible mechanism, the facilitation model, has suggested that information of
the input may have stored on the selective neurons that are specifically responsible
for the presented stimuli after the first encounter of processing. As a result,
less neural processing is required to respond to the same stimulus with
repetitions and less processing time is expected. The reduction in neural
processing could be attributed to shortened neural duty cycle in which the
strength of neuron firing is increased but with a shorter duration (Kar &
Krekelberg, 2016).

Previous studies have shown several mechanisms
under the fMRI adaptation paradigms, such as the reduction in firing rate
reflecting neural fatigue, sharpened neural responses with the inhabitation of
non-selective neurons, facilitated processing attributed to shortened neural
duty cycle; however, there are still many unknowns for the neural mechanisms of
fMRI adaptation and probably other possible mechanism that could explain the
adaptation effect.

CONCLUSION

The fMRI adaptation is a phenomenon where
repetitions of the same visual stimulus cause a consistent and gradual
reduction in activation in a given voxel that serves as a tool to study the
properties of neuronal populations on a sub-voxel scale. Specifically, not only
do recent studies have provided evidence that the fMRI signal recorded could reveal
neural selectivity under the fMRI adaptation paradigms, but also enabled the
investigation of the invariance of the neural responses from the neural
populations within the imaged voxels enhancing the functional resolution. The adaptation
effect reflected by a gradual reduction in BOLD signal with repeated
presentations across changes between stimuli demonstrates common neural populations
invariant to attributes being varied and the recovery responses from adaptation
shown by an increase in BOLD signal implies the selectivity of specific
stimulus properties among neuronal representations.

There are several mechanisms being proposed to
account for the reduction in neural responses observed after a stimulus being
presented repeatedly or with a prolonged time. Neural fatigue is one of the
most commonly proposed mechanisms that might induce the adaptation effect as a
result of a reduction in neuronal firing rate. Another model also suggests the
mismatch in the observed and expected stimuli could lead to the differences observed
in identical and varied conditions. A reduction in BOLD signal does not
necessarily indicate neural fatigue or a decline in performance but a
sharpening and facilitation model. The sharpening model explains the reduction
in BOLD signal as the suppression of non-selective neurons that are irrelevant
to the properties of the visual stimuli and therefore, less neurons are
involved in the processing; while the facilitation model proposes that the
reduction in BOLD signal could be due to the fact that information of the input
might have primed on the selective neurons that are specifically responsible
for the presented stimuli after the initial presentation of processing. Less
processing time and neural processing are required to respond to the same
stimulus which could be attributed to shortened neural duty cycle. The above
mentioned mechanisms that are responsible for inducing the adaptation effect
might aim to reduce cognitive and neural costs for irrelevant, known
information and allocate resources to the processing of relevant and newly
received information.

 

 

BIBLIOGRAPHY

Armony,
J. L., Aubé, W., Angulo-Perkins, A., Peretz, I., & Concha, L. (2015). The
specificity of neural responses to music
and their relation to voice processing: An fMRI-adaptation       study. Neuroscience letters, 593,
35-39.

Barron,
H. C., Garvert, M. M., & Behrens, T. E. (2016). Repetition suppression: a
means to      index neural representations
using BOLD?. Philosophical Transactions of the Royal       Society B: Biological Sciences, 371(1705).

Fukuda,
M., Poplawsky, A. J., & Kim, S. G. (2016). Submillimeter-resolution fMRI:
Toward   understanding local neural
processing. Progress in brain research, 225, 123-152.

Ganel,
T., Gonzalez, C. L., Valyear, K. F., Culham, J. C., Goodale, M. A., &
Köhler, S. (2006).        The relationship
between fMRI adaptation and repetition priming. Neuroimage, 32(3),
       1432-1440.

Grill-Spector,
K., & Malach, R. (2001). fMR-adaptation: a tool for studying the functional
      properties of human cortical
neurons. Acta psychologica, 107(1), 293-321.

Harris,
R. J., Young, A. W., & Andrews, T. J. (2014). Brain regions involved in
processing        facial identity and
expression are differentially selective for surface and edge information. NeuroImage, 97,
217-223.

Henry,
C. A., Joshi, S., Xing, D., Shapley, R. M., & Hawken, M. J. (2013).
Functional      characterization of the
extraclassical receptive field in macaque V1: contrast, orientation,      and temporal dynamics. Journal of
Neuroscience, 33(14), 6230-6242.

Kar,
K., & Krekelberg, B. (2016). Testing the assumptions underlying fMRI
adaptation using intracortical recordings
in area MT. Cortex, 80, 21-34.

Lai,
J., Pancaroglu, R., Oruc, I., Barton, J. J., & Davies-Thompson, J. (2014).
Neuroanatomic   correlates of the
feature-salience hierarchy in face processing: an fMRI-adaptation       study. Neuropsychologia, 53,
274-283.

Larsson,
J., & Smith, A. T. (2012). fMRI repetition suppression: neuronal adaptation
or    stimulus expectation?. Cerebral
Cortex, 22(3), 567-576.

Larsson,
J., Solomon, S. G., & Kohn, A. (2016). fMRI adaptation revisited. Cortex, 80,
   154-160.

 

Tsougos,
I. (2017). Advanced MR Neuroimaging: From Theory to Clinical Practice.
CRC         Press.

Van
Leeuwen, T. (2014). Color Specificity in the Human V4 Complex–An fMRI
Repetition    Suppression Study. In Advanced
brain neuroimaging topics in health and disease-methods
and applications. InTech.

Weibert,
K., Harris, R. J., Mitchell, A., Byrne, H., Young, A. W., & Andrews, T. J.
(2016). An image-invariant neural
response to familiar faces in the human medial temporal lobe.   Cortex, 84, 34-42.

Weiner,
K. S., Sayres, R., Vinberg, J., & Grill-Spector, K. (2010). fMRI-adaptation
and    category selectivity in human
ventral temporal cortex: regional differences across time scales. Journal of neurophysiology, 103(6),
3349-3365.

Xu,
Y., Turk-Browne, N. B., & Chun, M. M. (2007). Dissociating task performance
from fMRI repetition attenuation in
ventral visual cortex. Journal of
Neuroscience, 27(22), 5981–5985.

Zimmermann, E., Weidner, R., Abdollahi, R. O.,
& Fink, G. R. (2016). Spatiotopic adaptation         in visual areas. Journal of Neuroscience, 36(37),
9526-9534.

 

x

Hi!
I'm Brent!

Would you like to get a custom essay? How about receiving a customized one?

Check it out