Dietrich Samuel Schwarzkopf

Stochastic Resonance Effects Reveal the Neural Mechanisms of Transcranial Magnetic Stimulation

Transcranial magnetic stimulation (TMS) is a popular method for studying causal relationships between neural activity and behavior. However, its mode of action remains controversial, and so far there is no framework to explain its wide range of facilitatory and inhibitory behavioral effects. While some theoretical accounts suggest that TMS suppresses neuronal processing, other competing accounts propose that the effects of TMS result from the addition of noise to neuronal processing. Here we exploited the stochastic resonance phenomenon to distinguish these theoretical accounts and determine how TMS affects neuronal processing. Specifically, we showed that online TMS can induce stochastic resonance in the human brain. At low intensity, TMS facilitated the detection of weak motion signals, but with higher TMS intensities and stronger motion signals, we found only impairment in detection. These findings suggest that TMS acts by adding noise to neuronal processing, at least in an online TMS protocol. Importantly, such stochastic resonance effects may also explain why TMS parameters that under normal circumstances impair behavior can induce behavioral facilitations when the stimulated area is in an adapted or suppressed state.

Better ways to improve standards in brain-behavior correlation analysis.

Rousselet and Pernet (2012) demonstrate that outliers can skew Pearson correlation. They claim that this leads to widespread statistical errors by selecting and re-analyzing a cohort of published studies. However, they report neither the study identities nor inclusion criteria for this survey, so their claim cannot be independently replicated. Moreover, because their selection criteria are based on the authors’ belief that a study used misleading statistics, their study represents an example of “double dipping” (Kriegeskorte et al., 2009). The strong claims they make about the literature are therefore circular and unjustified by their data. Their purely statistical approach also does not consider the biological context of what observations constitute outliers. In discussion, they propose that the skipped correlation (Wilcox, 2005) is an appropriate alternative to the Pearson correlation that is robust to outliers. However, this test lacks statistical power to detect true relationships (Figure ​(Figure1A)1A) and is highly prone to false positives (Figure ​(Figure1B).1B). These factors conspire to drastically reduce the sensitivity of this test in comparison to other procedures (Appendix 1). Further, it is susceptible to the parameters chosen for the minimum covariance estimator to identify outliers but these parameters are not reported. Figure 1 Statistical power (A) and false positive rates (B) for four statistical tests and four sample sizes based on 10,000 simulations (see Appendix 1 for details). Outliers can drastically inflate false positives for Pearson correlation (note the difference ... Their argument fails to consider a broad literature on robust statistics, although an extensive review is outside the scope of this commentary. We limit ourselves instead to presenting a practical alternative to their approach: Shepherd’s pi correlation (http://www.fil.ion.ucl.ac.uk/~sschwarz/Shepherd.zip). We identify outliers by bootstrapping the Mahalanobis distance, Ds, of each observation from the bivariate mean and excluding all points whose average Ds is 6 or greater. Shepherd’s pi is Spearman’s rho but the p-statistic is doubled to account for outlier removal (Appendix 2). This compares very well in power (Figure ​(Figure1A)1A) to other tests and is more robust to the presence of influential outliers (Figure ​(Figure1B).1B). We replot the data Rousselet and Pernet presented in their Figure 2. The conclusions drawn from Shepherd’s pi are comparable to skipped correlation but less strict in situations where a relationship is likely (Figure ​(Figure1C,1C, Figures ​FiguresA1A1 and ​andA2A2 in Appendix). Consider for instance the data in Figure ​Figure1C-1.1C-1. Pearson and Spearman correlation applied to these data are comparable. This implies that the assumptions of Pearson’s r were probably met in this case. The skipped correlation (r’) does not reach significance but nevertheless shows a similar relationship, consistent with our demonstration above that it is too conservative a measure. Under Shepherd’s pi, however, the relationship between these variables is significant. Indeed, reflecting our intimate knowledge of these data (Schwarzkopf et al., 2011), we already know that the relationship studied here replicates for separate behavioral measures (see Schwarzkopf et al., 2011 SOM). A similar pattern was observed for other data, e.g., Figure ​Figure1C-2.1C-2. In some cases skipped correlation even removes the majority of data as outliers (e.g., their Figure 2E), which borders on the absurd. Rousselet and Pernet also claim that none of the studies that they surveyed considered the correlation coefficient and its confidence intervals. Cohen defined that 0.3 0.5, that is, at least 25% of the variance is explained. A correlation accounting for ~15% of variance is thus not particularly “modest” as they state. Naturally, this taxonomy is somewhat arbitrary but when relating complex cognitive functions to brain measures we are unlikely to find very high r, except for trivial relationships (Yarkoni, 2009). Their failure to find reported confidence intervals in the literature is also puzzling because it does not accurately report the published work they considered. For example, our study, reproduced in their Figure 2A, reported bootstrapped 95% confidence intervals in the figure (Schwarzkopf et al., 2011). They also do not consider important aspects of what confidence intervals reflect. Naturally, a confidence interval is an indicator of the certainty with which the effect size can be estimated. However, it depends on three factors: the strength of the correlation, the sample size, and the data distribution. Because Pearson correlation assumes a Gaussian distribution we can predict the confidence interval for any given r. If the bootstrapped confidence interval differs from this prediction, the data probably do not meet the assumptions. Rousselet and Pernet’s example for bivariate outliers (their Figure 1D) illustrates this: the predicted confidence interval for r = 0.49 with n = 17 should be (0.01, 0.79). However, the bootstrapped confidence interval for this example is (−0.19, 0.87), much wider and also overlapping zero. This indicates that outliers skew the correlation and that it should not be considered significant. Compare this to Figure ​Figure1C-11C-1 (their Figure 2A): the nominal confidence interval should be (−0.65, −0.02); the actual bootstrapped interval is very similar: (−0.67, −0.03). Therefore, the use of Pearson/Spearman correlation was justified here. We propose simple guidelines to follow when testing correlations. First, use Spearman’s rho because it captures non-linear relationships. Second, bootstrap confidence intervals. Third, if the interval differs from the nominal interval, apply Shepherd’s pi as a more robust test. Fourth, estimate the reliability of individual observations, especially in cases where outliers strongly affect results. Outliers are frequently the result of artifacts or measurement error. Our last point highlights an important general concern we have with Rousselet and Pernet’s argument. Statistical tests are important tools to be used by researchers for interpreting their data. However, the goal of neuroscience is to answer biologically relevant questions, not to produce statistically significant results. No statistical procedure can determine whether a biological question is valid or if a theory is sound. Rather, one has to inspect each finding and each data point in its own right, evaluating the data quality and the potential confounds on a case-by-case basis. Outliers should not be determined solely by statistical tests but must take into account biological interpretation (Bertolino, 2011; Schott and Duzel, 2011). And finally, there is only one way any finding can be considered truly significant; when upon repeated replication it passes the test of time.

The frequency of visually induced gamma-band oscillations depends on the size of early human visual cortex

The structural and functional architecture of the human brain is characterized by considerable variability, which has consequences for visual perception. However, the neurophysiological events mediating the relationship between interindividual differences in cortical surface area and visual perception have, until now, remained unknown. Here, we show that the retinotopically defined surface areas of central V1 and V2 are correlated with the peak frequency of visually induced oscillations in the gamma band, as measured with magnetoencephalography. Gamma-band oscillations are thought to play an important role in visual processing. We propose that individual differences in macroscopic gamma frequency may be attributed to interindividual variability in the microscopic architecture of visual cortex.

Variability in visual cortex size reflects tradeoff between local orientation sensitivity and global orientation modulation

The surface area of early visual cortices varies several fold across healthy adult humans and is genetically heritable. But the functional consequences of this anatomical variability are still largely unexplored. Here we show that interindividual variability in human visual cortical surface area reflects a tradeoff between sensitivity to visual details and susceptibility to visual context. Specifically, individuals with larger primary visual cortices can discriminate finer orientation differences, whereas individuals with smaller primary visual cortices experience stronger perceptual modulation by global orientation contexts. This anatomically correlated tradeoff between discrimination sensitivity and contextual modulation of orientation perception, however, does not generalize to contrast perception or luminance perception. Neural field simulations based on a scaling of intracortical circuits reproduce our empirical observations. Together our findings reveal a feature-specific shift in the scope of visual perception from context-oriented to detail-oriented with increased visual cortical surface area.

Exploring the parahippocampal cortex response to high and low spatial frequency spaces

The posterior parahippocampal cortex (PHC) supports a range of cognitive functions, in particular scene processing. However, it has recently been suggested that PHC engagement during functional MRI simply reflects the representation of three-dimensional local space. If so, PHC should respond to space in the absence of scenes, geometric layout, objects or contextual associations. It has also been reported that PHC activation may be influenced by low-level visual properties of stimuli such as spatial frequency. Here, we tested whether PHC was responsive to the mere sense of space in highly simplified stimuli, and whether this was affected by their spatial frequency distribution. Participants were scanned using functional MRI while viewing depictions of simple three-dimensional space, and matched control stimuli that did not depict a space. Half the stimuli were low-pass filtered to ascertain the impact of spatial frequency. We observed a significant interaction between space and spatial frequency in bilateral PHC. Specifically, stimuli depicting space (more than nonspatial stimuli) engaged the right PHC when they featured high spatial frequencies. In contrast, the interaction in the left PHC did not show a preferential response to space. We conclude that a simple depiction of three-dimensional space that is devoid of objects, scene layouts or contextual associations is sufficient to robustly engage the right PHC, at least when high spatial frequencies are present. We suggest that coding for the presence of space may be a core function of PHC, and could explain its engagement in a range of tasks, including scene processing, where space is always present.

Early visual learning induces long-lasting connectivity changes during rest in the human brain

Spontaneous fluctuations in resting state activity can change in response to experience-dependent plasticity and learning. Visual learning is fast and can be elicited in an MRI scanner. Here, we showed that a random dot motion coherence task can be learned within one training session. While the task activated primarily visual and parietal brain areas, learning related changes in neural activity were observed in the hippocampus. Crucially, even this rapid learning affected resting state dynamics both immediately after the learning and 24 h later. Specifically, the hippocampus changed its coupling with the striatum, in a way that was best explained as a consolidation of early learning related changes. Our findings suggest that long-lasting changes in neuronal coupling are accompanied by changes in resting state activity.

Differing causal roles for lateral occipital cortex and occipital face area in invariant shape recognition.

The human extrastriate visual cortex contains functionally distinct regions where neuronal populations exhibit signals that are selective for objects. How such regions might play a causal role in underpinning our ability to recognize objects across different viewpoints remains uncertain. Here, we tested whether two extrastriate areas, the lateral occipital (LO) region and occipital face area (OFA), contained neuronal populations that play a causal role in recognizing two‐dimensional shapes across different rotations. We used visual priming to modulate the rotation‐sensitive activity of neuronal populations in these areas. State‐dependent transcranial magnetic stimulation (TMS) was applied after the presentation of a shape and immediately before a subsequent probe shape to which participants had to respond. We found that TMS applied to both the LO region and OFA modulated rotation‐invariant shape priming but, whereas the LO region was modulated by TMS for small rotations, the OFA was modulated for larger rotations. Importantly, our results demonstrate that a node in the face‐sensitive network, the OFA, participates in causally relevant encoding of non‐face stimuli.

Knowing with Which Eye We See: Utrocular Discrimination and Eye-Specific Signals in Human Visual Cortex

Neurophysiological and behavioral reports converge to suggest that monocular neurons in the primary visual cortex are biased toward low spatial frequencies, while binocular neurons favor high spatial frequencies. Here we tested this hypothesis with functional magnetic resonance imaging (fMRI). Human participants viewed flickering gratings at one of two spatial frequencies presented to either the left or the right eye, and judged which of the two eyes was being stimulated (utrocular discrimination). Using multivoxel pattern analysis we found that local spatial patterns of signals in primary visual cortex (V1) allowed successful decoding of the eye-of-origin. Decoding was above chance for low but not high spatial frequencies, confirming the presence of a bias reported by animal studies in human visual cortex. Behaviorally, we found that reliable judgment of the eye-of-origin did not depend on spatial frequency. We further analyzed the mean response in visual cortex to our stimuli and revealed a weak difference between left and right eye stimulation. Our results are thus consistent with the interpretation that participants use overall levels of neural activity in visual cortex, perhaps arising due to local luminance differences, to judge the eye-of-origin. Taken together, we show that it is possible to decode eye-specific voxel pattern information in visual cortex but, at least in healthy participants with normal binocular vision, these patterns are unrelated to awareness of which eye is being stimulated.

Effective Connectivity within Human Primary Visual Cortex Predicts Interindividual Diversity in Illusory Perception

Visual perception depends strongly on spatial context. A classic example is the tilt illusion where the perceived orientation of a central stimulus differs from its physical orientation when surrounded by tilted spatial contexts. Here we show that such contextual modulation of orientation perception exhibits trait-like interindividual diversity that correlates with interindividual differences in effective connectivity within human primary visual cortex. We found that the degree to which spatial contexts induced illusory orientation perception, namely, the magnitude of the tilt illusion, varied across healthy human adults in a trait-like fashion independent of stimulus size or contrast. Parallel to contextual modulation of orientation perception, the presence of spatial contexts affected effective connectivity within human primary visual cortex between peripheral and foveal representations that responded to spatial context and central stimulus, respectively. Importantly, this effective connectivity from peripheral to foveal primary visual cortex correlated with interindividual differences in the magnitude of the tilt illusion. Moreover, this correlation with illusion perception was observed for effective connectivity under tilted contextual stimulation but not for that under iso-oriented contextual stimulation, suggesting that it reflected the impact of orientation-dependent intra-areal connections. Our findings revealed an interindividual correlation between intra-areal connectivity within primary visual cortex and contextual influence on orientation perception. This neurophysiological-perceptual link provides empirical evidence for theoretical proposals that intra-areal connections in early visual cortices are involved in contextual modulation of visual perception.

Neural Population Tuning Links Visual Cortical Anatomy to Human Visual Perception

Summary The anatomy of cerebral cortex is characterized by two genetically independent variables, cortical thickness and cortical surface area, that jointly determine cortical volume. It remains unclear how cortical anatomy might influence neural response properties and whether such influences would have behavioral consequences. Here, we report that thickness and surface area of human early visual cortices exert opposite influences on neural population tuning with behavioral consequences for perceptual acuity. We found that visual cortical thickness correlated negatively with the sharpness of neural population tuning and the accuracy of perceptual discrimination at different visual field positions. In contrast, visual cortical surface area correlated positively with neural population tuning sharpness and perceptual discrimination accuracy. Our findings reveal a central role for neural population tuning in linking visual cortical anatomy to visual perception and suggest that a perceptually advantageous visual cortex is a thinned one with an enlarged surface area.