Lateral Modulation of BOLD Activation in Unstimulated Regions of the Human Visual Cortex

 

 

Chien-Chung Chen*&, Christopher W. Tyler*#, Chia-Li Liu+& & Yao-Hong Wang+

*The Smith-Kettlewell Eye Research Institute, +The MRI Laboratory, National Taiwan University, &Department of Psychology, National Taiwan University


Abstract

After staring a blank region surrounded by a dynamic background for a few seconds, observers report a twinkle aftereffect in the blank region after the dynamic surround is removed. The significance of this twinkle aftereffect is that it occurs at a location that received no stimulation. Based on this aftereffect, we measured the blood oxygenation level dependent (BOLD) activation in the primary visual cortex while the observers were viewing a flickering pin-wheel pattern consisting of checkerboard wedges alternating with a blank test with a 42s period. Retinotopic regions corresponding both to the wedges and to the inter-wedge regions in the pin-wheel pattern showed activation highly correlated with the test sequence. In the inter-wedge regions the correlation was negative. While the BOLD activation in the visual cortex is generally considered to be retinotopically driven by the visual stimuli, we were able to show a sustained negative activation in the unstimulated regions, with properties that correspond to those of the perceived aftereffect. The results further demonstrate that the effect of inhibition of cortical activity is to reduce the BOLD activity level in the cortex.

Text Box: Figure 1. Stimuli used in the experiment. The adapter (a) was a pin-wheel pattern consist of 4 checkerboard wedges counter-phase flickering at 10Hz (1Hz in the control experiment). The adapter was alternated with a blank test (b) that contained only a fixation point. All observers reported twinkling afterimage (illustrated in (c) in the inter-wedge regions of the adapter while viewing the blank test. However, unlike the filling-in effect where the checkerboard pattern is clearly visible in the inter-wedge region, in the twinkle aftereffect, the detail spatial structure of the afterimage is not always clear to the observers. The vast majority of functional magnetic resonance imaging (fMRI) studies measure the blood oxygenation level dependent (BOLD) activation that implies neuronal activity changes. Here, we examine two generally accepted assumptions of BOLD activation in the visual cortex. First, based on the fact that BOLD activation in the visual cortex is retinotopic1,2, the activation of a certain voxel is considered to be driven by a visual stimulus that projects onto its corresponding retina location while voxels corresponding to unstimulated retinal locations are devoid of activation change. Second, in a block design study (i.e., BOLD changes observed between a stimulated and a control condition) the BOLD activation is positively correlated with the stimulus onset. It has been shown that the BOLD response to a brief stimulus consists of a stimulus dependent positive activation followed by a post-stimulation undershoot3. Hence, the BOLD activation should increase when the stimulus is presented and decrease if not. Here, we report a novel effect in functional magnetic resonance techniques that reveal a sustained negative BOLD signal in the unstimulated regions which is correlated with human visual percept and due to the operation of long-range neural interactions rather than the simple blood dynamics. This, in turn, gives light the problem in previous report4 of sustain negative BOLD activation, which cannot determine the cause to be the inhibition of neural or blood drains from inactivated regions of cortex to activated regions, generating a negative BOLD signal in the absence of neural inhibition.

The experiment is based on known lateral interactions involved in the twinkle aftereffect. After staring at a grey region surrounded by a dynamic patterned background (adapter), an observer will perceive a twinkling afterimage in the location of the grey region when the pattern stimulus is removed. 5,6 That is, the aftereffect is induced in a region that had never received any stimulation in either the adapting and the test phase and must result from the a lateral interaction between the mechanisms responding to the dynamic surround region and the blank test region. Our experiment used a block design in which the 21s adapter alternated with a uniform 21s test over a 42s period. Two blank epochs were extended to 42s at the middle and the end of the experiment to establish the activation baseline. The adapting stimulus was a pinwheel pattern consisting of 4 counter-phase flickering (10Hz) checkerboard wedges, each extending p/4 radians and separated from each other by p/4 (see Fig. 1). All observers verbally reported twinkle aftereffects in the inter-wedge region when viewing the uniform test field following the 21s adaptation period. We used a 3T scanner to record the BOLD activation of six observers.

Text Box: Figure 2. A typical activation map for the experiment on a 3D rendered surface of the left (L) and right (R) occipital cortex. Coloured patches denote voxels showing significant activation change (|r| >0.35) in the experiment (for this observer, 18.1% of all voxels identified by the spatial localizer). Green: activated voxels in the stimulated wedge regions; blue: voxels in the inter-wedge with activation negatively correlated with the experimental sequence; and yellow: voxels in the inter-wedge with positively correlated activation. For this observer, 66% of the significantly activated inter-wedge voxels showed a negative correlation with the experimental sequence (blue). The white dots mark the calcarine sulcus and the black open circles the occipital pole. A spatial localizer (see Methods) identified the cortical regions corresponding retinotopically to the wedge and inter-wedge regions in the adapter. The typical behavior depicted in Fig. 2 illustrates that a substantial proportion of the inter-wedge voxels showed a significant BOLD activation change (correlation coefficient |r| >0.35) between the adapting and the test epochs, even though those voxels corresponded to the retinal areas that were never stimulated during the experiment. The majority (60-80%) of the unstimulated activation was in counter-phase with the adapter onset (blue patches in Fig. 2): the BOLD signal decreased when the adapter appeared and increased following its disappearance. A minority of the unstimulated activation was in phase with the stimulus onset (yellow patches in Fig. 2), presumably corresponding to spread of the activation from the stimulated regions. The wedge-region voxels that showed significant activation changes in the experiment are denoted as green spots in Fig. 2 as reference points. It can be seen that all the yellow patches lie between the green stimulated and the blue counterphase regions, consistent with the spread of activation hypothesis.

Text Box: Figure 3. (a) Averaged percentage change of BOLD activation time series for interwedge voxels showing negative correlation with the stimulation periods (grey bars) relative to the blank screen periods (white bars) and (b) is the best fit impulse response function to this time series. (c) Averaged percentage change of BOLD activation time series for interwedge voxels shown negative correlation with the stimulation periods The error bars denote 1 standard error of the means. The smooth curve is the fit of a convolution of the experimental sequences with a difference-of-Gamma impulse response function: where wi, ai, and bi, i=1, 2 are free parameters. (e)-(i) The impulse response functions for counter-phase voxels (solid curves), and in-phase voxels (dashed curves). The in-phase curves were plotted upside down to give a better comparison of the two functions. An analysis of the time series reveals both the suppression and rebound of BOLD activation in the counter-phase inter-wedge voxels. The BOLD activation decreased with the onset of the adapter and increased with the offset (Fig 3a), contrast to the wedge region voxels (Fig 3c) where activation increased with the onset of the adapter. The activation returns to a base level during the two extended blank periods for the in-phase voxels but not counter-phase voxels, suggesting a difference in time course of activation. To analyze this difference, we fit the activation time series by convolving the experimental stimulation sequence with a model impulse response function defined as the difference of two Gamma functions. The parameters of the two Gamma functions were optimized separately for each observer for best fit to the respective data sets. For the counter-phase voxels, five out of the six observers showed a triphasic impulse response (solid curves, Fig. 3b) with an early reduction followed by an increase and then another dip, although the sixth observer did not show the second dip. The in-phase voxels in the stimulated regions showed a biphasic response function, beginning with an increase followed by a decrease. Comparison of the two sets of impulse-response parameters shows that half-height full width of the positive portion of the counter-phase response (solid curves in Fig. 3e-i) is significantly smaller (t(5)=2.39, p=0.0311 <0.05) than that of the undershoot portion of the in-phase response (dashed curves in Fig 3e-i, plotted upside down for ready comparison). This discrepancy suggested that the counter-phase activation is caused by a different hemodynamic mechanism from the in-phase activation. The modulation of the inter-wedge activation cannot be explained as a result of strong activity in the wedge regions draining blood from the unstimulated regions (blood drain). If this was the case, the interwedge activation would have a similar time course as that of the wedge regions

Text Box: Figure 4. An activation map for the 1 Hz activation experiment. Coloured spots denote voxels showing significant activation change (|r| >0.35) in the experiment (for this observer, 20.6% of all voxels identified by the spatial localizer). The colour codes are the same as in Figure 2. For this observer, the inter-wedge voxels showing significantly negative correlation with the experimental sequence dropped from 66% in the 10 Hz adapter experiment to 48%. Hence, the blue spots occupy much less space on this 3-D render image of occipital cortex.   

As a further control for the hemodynamic compensation hypothesis, we changed the flicker frequency of the counter-phase alternation from 10Hz to 1Hz. If the activity modulation in the unstimulated regions is due to the blood draining from those regions, we should expect the activation pattern with the 1Hz adapter to be similar to that for the 10Hz adapter, with a substantial proportion of the inter-wedge voxels being in counterphase with the adapter onset. On the other hand, it is known that 1Hz dynamic surround induces little or no twinkle aftereffect but strong filling-in during adaptation in the unstimulated regions6. Unlike the blood-draining hypothesis predicts that there should be about the same proportion of inter-wedge voxels with counter-phase activation pattern relative to in-phase voxels as at 10 Hz, the aftereffect hypothesis predicts far fewer counterphase voxels at 1 Hz. The experimental results with the 1Hz adapter clearly showed that, in the interwedge regions, the proportion of the in-phase voxels (50~70%) outweighed counter-phase voxels in the inter-wedge region (Fig. 4).

 

Two possible interpretation of the twinkle aftereffect have been proposed. The first interpretation is based on the perceptual filling-in observed in artificially induced scotoma experiments, in which observers often reported that the blank test region vanishes and is replaced by the pattern of the surround after viewing the stimulus for as little as 5s5. This filling-in process may persist upon cessation of stimulation to produce the twinkle aftereffect. The second interpretation suggests that the activation of the visual mechanisms responding to the surround during adaptation produces a suppressive effect on the mechanisms in the unstimulated test regions6. This suppression in turn causes a post-stimulus rebound in the test regions.

The counter-phase activation pattern is inconsistent with the filling-in hypothesis. It is known that filling-in occurs in less than 5s5,7 after the adapter onset. Given the much longer duration of the adaptation period, we should expect the inter-wedge activation, after allowing a small delay, to be almost in-phase with the adapter. This relation is inconsistent with the activation of the majority of the inter-wedge voxels, which is counterphase to the stimulus onset. On the other hand, the lateral suppression hypothesis predicts the unstimulated activation to be suppressed during the adaptation and to rebound during blank test. This prediction is consistent with our results.

In conclusion, we have demonstrated a lateral rebound of BOLD activation that correlates with the human percept of activity in unstimulated regions following stimulation of their neighboring regions. This result was not due to blood flow from unstimulated to stimulated regions during adaptation. Our results indicate that the BOLD signal can be reduced by inhibitory influences on a region of cortex. It also rejects the filling-in hypothesis of the twinkle aftereffect.

Methods

Stimulus and the experimental design. The adapter contained four 45o checkerboard wedges separated from each other by 45o blank regions set at the mean luminance and extended 6o from the central fovea. The checkerboards were counterphase flickering at 10 Hz during the presentation. In the main experiment, the adapter alternated with a blank field, which contained only a fixation mark, in a 42s period, with six periods in each scan session. In addition, a 21s blank period was inserted in the middle and at the end of the session to establish the baseline unstimulated level. The control experiment was the same as the main experiment except that the flicker rate of the checkerboards was reduced to 1 Hz. The cortical regions retinotopically corresponding to the stimulated wedges and to the inter-wedge regions were determined by a spatial localizer. The spatial localizer contained two patterns: one being the same as the adapter and the other being the same as the adapter but rotated clockwise by 45o. The two patterns alternated with each other in a 42s period for six periods. Both patterns contained checkerboards counter phase flickering at 10Hz. Five observers with normal vision (three females, two males) participated in the study, including four volunteers and one of the authors.

Data acquisition and analysis. The images were collected with a Bruker 3T scanner located at National Taiwan University. A high-resolution anatomical (T1-weighted) MRI volume scan of the entire head was run once on each observer (voxel size = 1 x 1 x 0.75 mm). Within each scanning session, both functional (T2*-weighted, BOLD) responses and anatomical (T1-weighted) images were acquired in identical planes. The images were collected in 18 transverse planes parallel to the AC-PC (anterior commissure Ð posterior commissure) line. An Echo-planar imaging sequence8 was used to acquired the functional data (TR = 3500ms, TE = 35ms, flip angle = 90o, voxel resolution = 2.34 x 2.34 x 3mm). The main experiment lasted 304.5s (87 images). The first 10.5s (3 images) was excluded from further analyses. Thus, the data analyzed for each scan spanned 294s (84 images). We used SPM9 to realign the EPI images acquired in the spatial location, the main and the control experiment. The EPI images were coregistered with the T1-weighted image with a public available software10, which also performed part of the data analysis and 3D rendering of Figs. 1 & 3. Statistic analysis of the BOLD activation was based on linear correlation between the time series and the experimental sequence1.

 

Acknowledgement.

This study was partially supported by NIH NEI 13025 to CWT. CLL thanks the assistance of Dr. K. C. Liang of the Psychology Department, National Taiwan University through funding from the National Science Council, Taiwan. The equipment was partially funded by an Interdisciplinary Equipment Grant of National Taiwan University to Dr. K. C. Liang.

 

References

1. Engel, S. A., Glover, G. H., Wandell, B. A. Retinotopic organization in human visual cortex and the spatial precision of functional MRI. Cereb. Cortex 7, 181-92. 1997.

2. Tootell, R. B., Mendola JD, Hadjikhani NK, Ledden PJ, Liu AK, Reppas JB, Sereno MI, Dale AM. Functional analysis of V3A and related areas in human visual cortex. J Neurosci 17, 7060-78. 1997.

3. Buxton RB, Wong EC, Frank LR. Dynamics of blood flow and oxygenation changes during brain activation: the balloon model. Magn Reson Med. 39, 855-64. 1998.

4. Shmuel A., Yacoub, E., Pfeuffer, J., Van de Moortele, P. F., Adriany, G., Hu, X., and Ugurbil, K. Sustained negative BOLD, blood flow and oxygen consumption response and its coupling to the positive response in the human brain. Neuron 36, 1195-210. 2002.

5. Ramachandran, V. S. & Gregory, R. L. Perceptual filling in of artificially induced scotomas in human vision. Nature 350, 699-702. 1991.

6. Hardage, L. & Tyler, C. W. Induced twinkle aftereffect as a probe of dynamic visual processing mechanisms. Vision Res. 35, 757-66. 1995.

7. Ramachandran, V. S., Gregory, R. L. & Aiken, W. Perceptual fading of visual texture borders. Vision Re.s 33, 717-27. 1993.

8. Stehling, M. K., Turner, R., Mansfield, P. Echo-planar imaging: magnetic resonance imaging in a fraction of a second. Science 254, 43-50. 1991.

9. Friston, K. J., Holmes, A. P., Worsley, K. J., Poline, J. P., Frith, C. D. & Frackowiak, R. S. J. Statistical parametric maps in functional imaging: a general linear approach. Human Brain Mapping 2,189-210. 1995.

10. Wandell, B.A., Chial, S. & Backus, B.T. Visualization and measurement of the cortical surface. Journal of Cognitive Neuroscience 12, 739-52. 2000.