Our "Neuronal Basis of Fear Generalization" manuscript has been accepted to be published in Nature Neuroscience.
You can download the pdf here.
It has been also highlighted in Nature Reviews Neuroscience.
Effect of aversive learning on discrimination of faces
In her Msc thesis, Lea Kampermann shows that humans can perceptually discriminate faces better, when these are paired with an aversive outcome. This effect was specific to the face, which was paired with an aversive outcome and was not observed for the one which was kept neutral throughout the experiment. Furthermore the effect was strongest when these faces were presented at shorter durations (~.6 s) allowing participants to make no more than two fixations per trial.
Her thesis contains also a detailed account on the methodology for generating face-stimuli that are perceptually calibrated to form a two-dimensional similarity gradient with equal perceptual steps between faces. The methodology is an extension of work from Yue et al. (Vision Research, 2012). If you wish to use these stimuli for your experiment they are available upon request.
Her thesis contains also a detailed account on the methodology for generating face-stimuli that are perceptually calibrated to form a two-dimensional similarity gradient with equal perceptual steps between faces. The methodology is an extension of work from Yue et al. (Vision Research, 2012). If you wish to use these stimuli for your experiment they are available upon request.
 |
| Perceptually calibrated set of faces according to a simple primary visual cortex forming a circular similarity gradient. Details on their production can be read in Msc Thesis of Lea (please contact any of us for a pdf) . |
Categorical, yet graded--single-image activation profiles of human category-selective cortical regions.
Mur M et al. investigated the selectivity of activity levels in parahippocampal place area (PPA) and fusiform face area (FFA) evoked by single images. They focus here only on the average BOLD activity within carefully selected ROIs.
The paper is very creative in terms of new analyses methods, relies heavily on rank orders and hypothesis testing with bootstrapping.
First it establishes the fact that PPA and FFA behaves as expected, that is face stimuli for FFA and place stimuli for PPA rank highest in terms of evoked activity. Overall PPA responses are more selective than FFA responses, reaching AUC values of 1 in both hemispheres. This results from the fact that faces evokes really high activity levels in the FFA, whereas, in the case of PPA inactivation by faces contribute to the PPA selectivity.
The rest of the report focuses on characterizing the category selectivity of these areas.
If an area is category selective in an ideal sense, non-preferred stimuli should never evoke higher activity levels than any other preferred stimulus, and if so, then only by chance due to noise.
The number of inverted pairs measures exactly the number of times one could identify violation of this rule by counting the number of times a stimulus from outside the category is ranked higher than a stimulus from within category. If these inverted pairs survive across multiple sessions (as measured by PRIP metric), this would be an evidence against ideal category selectivity. However as such, PRIP is not a very sensitive metric. For example one single preferred stimulus failing by chance to evoke any activity at all would be sufficient to generate very many inverted pairs, thus the metric seems to fluctuate highly non-linearly with respect to distance between inverted pairs. Therefore the authors, used the first sessions to identify preferred-nonpreferred pairs with largest activity difference, with the idea that an inversion with the largest activity difference would be the observation with least chance level. If these pairs survive across sessions, the difference in activity would then decrease only marginally and remain positive, thus providing evidence for stable inversions (as such, however this measure is also influenced by the noise on both the preferred and non-preferred stimulus). These analyses provide supporting evidence that FFA and PPA behave like an ideal category selective area, with the exception of left FFA, in line with the fact that left FFA is the ROI where smallest AUC values were observed (only though at ROI size of 128 voxels).
The paper is very creative in terms of new analyses methods, relies heavily on rank orders and hypothesis testing with bootstrapping.
First it establishes the fact that PPA and FFA behaves as expected, that is face stimuli for FFA and place stimuli for PPA rank highest in terms of evoked activity. Overall PPA responses are more selective than FFA responses, reaching AUC values of 1 in both hemispheres. This results from the fact that faces evokes really high activity levels in the FFA, whereas, in the case of PPA inactivation by faces contribute to the PPA selectivity.
The rest of the report focuses on characterizing the category selectivity of these areas.
If an area is category selective in an ideal sense, non-preferred stimuli should never evoke higher activity levels than any other preferred stimulus, and if so, then only by chance due to noise.
The number of inverted pairs measures exactly the number of times one could identify violation of this rule by counting the number of times a stimulus from outside the category is ranked higher than a stimulus from within category. If these inverted pairs survive across multiple sessions (as measured by PRIP metric), this would be an evidence against ideal category selectivity. However as such, PRIP is not a very sensitive metric. For example one single preferred stimulus failing by chance to evoke any activity at all would be sufficient to generate very many inverted pairs, thus the metric seems to fluctuate highly non-linearly with respect to distance between inverted pairs. Therefore the authors, used the first sessions to identify preferred-nonpreferred pairs with largest activity difference, with the idea that an inversion with the largest activity difference would be the observation with least chance level. If these pairs survive across sessions, the difference in activity would then decrease only marginally and remain positive, thus providing evidence for stable inversions (as such, however this measure is also influenced by the noise on both the preferred and non-preferred stimulus). These analyses provide supporting evidence that FFA and PPA behave like an ideal category selective area, with the exception of left FFA, in line with the fact that left FFA is the ROI where smallest AUC values were observed (only though at ROI size of 128 voxels).
Advanced Numerical Methods in Neuroscience Lecture is now online
Advanced Numerical Methods in Neuroscience is a lecture I am holding in the graduate school Neurodapt. You can reach the overview of the lecture from this link and download the course material from this link.
Categorical Representation of Visual Stimuli in the Primate Prefrontal Cortex
Freedman et al. generated pictures of complex objects using parametric combinations of 6 base images. These base images represented different kinds of felines or dogs, therefore their combinations gave rise to images that were graded in their category membership (e.g. whereas some combinations were clearly dog- or feline-like, others pictures were somewhere in between) while guaranteeing diversity within each category. These images were shown in a delayed-match-to-category task to monkeys. Solving this task requires a level of abstraction from the sheer appearances of the stimuli. Even at category boundaries where the discrimination is most difficult, the performance was high. They recorded activity of single neurons from prefrontal cortex, more precisely from the ventral part of the principal sulcus. Their results show evidence for neurons that are able to distinguish between these two, supposedly learnt categories. That is the responses are 1/ not gradual as the stimuli and 2/ characterized by a sharp step-like function at the category boundary. The data is clear, the interpretation is inline with the data. I find it unfortunately that pictures which are at the category boundary were not presented. And the study would gain if the similarity measure was in the perceptual space rather than in the stimulus domain. The 6 base images could in principle be tested on humans using perceptual mapping techniques.
OHBM Hamburg 2014 Abstract: Precision of Neuronal Representations during Fear Generalization
Precision of Neuronal Representations during Fear Generalization
Onat S., Büchel C.
Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Germany
Introduction
Fear generalization is usually conceived as resulting from a lack of precision in the neuronal representations about the aversive stimulus. Therefore the lack of certainty is believed to be the source of generalization that is observed on the behavioral level. Alternatively, fear generalization might be the result of an active neuronal process, whereby the nervous system tries to optimally control behavior based on what is already known. This scenario predicts that hyper-precise neuronal representations about the aversive stimulus would co-exist with a broader behavioral tuning. Using an event-related fMRI paradigm, we analyzed the precision of neuronal signals in different brain regions and compared these with different behavioral measurements.
Methods
We created 8 computer generated faces that were organized along a circular similarity gradient (Fig. 1, dashed line). A maximum-likelihood based multi-dimensional scaling method (Maloney et al., 2003) was used to confirm the circularity of the perceptual organization of these stimuli (Fig. 1, solid line). This gradual change in stimulus similarity was translated to an aversiveness gradient using a classical Pavlovian conditioning paradigm. To this end, one randomly selected face (CS+) was partially associated with an aversive electric shock. The most dissimilar face was kept as neutral (CS-). BOLD responses were recorded before and after the conditioning phase together with changes in skin conductance, as well as aversiveness ratings (n = 29).
Results
We identified a set of neuronal clusters that were significantly modulated as a function of increasing dissimilarity to the CS+ face. The average amplitude of evoked responses by the CS+, CS- and all intermediate faces is shown in Fig. 2 (mean ± SEM, red for CS+, cyan for CS-) for two clusters located in hippocampus and insula. The effect of conditioning is clearly seen as a modulation of responses centered on the CS+ face following the conditioning (bottom panels). These responses were fit with a Gaussian function, yielding parameterized fear-tuning profiles (Fig. 2, black curves), where alpha (𝛼) and sigma (𝜎) parameters characterized the strength and the width of the tuning profiles, respectively (Fig. 2).
Almost all clusters within this identified fear generalization network including a set of prefrontal, cingular, hippocampal, and face selective sensory sites exhibited a strong deactivation in response to CS+ face (Fig. 3, left panel). The right insula was the sole exception to this pattern (p < 0.001), showing a fear-tuning profile that was characterized by a net activation (Fig. 3, left panel, top bar). Among all the clusters investigated, the insula showed the sharpest fear tuning (Fig. 3, right panel, bottom bar). We next compared, the precision of insular aversive tuning to the fear tuning of skin conductance and aversiveness ratings. The width of insular tuning was even sharper than the tuning of any behavioral measure i.e. ratings (t(28)= -2.67, p = 0.0062) and skin conductance (t(28)= -2.23, p = 0.017) responses (Fig. 4).
Conclusion
These results show that the representation of the aversive stimulus is present in a hyper-precise manner within the neuronal networks responsible for fear-generalization. The imprecision of the tuning that is observed in other neuronal sites and at the behavior level seems to be mediated by a mechanism that actively “blurs” the source of the aversive event, rather than resulting from a lack of precision in the neuronal representations. Our results therefore suggest that a controlled imprecision rather than an imprecision in the control, is responsible for the generalization of fear in the healthy humans.
SFN 2013: Investigation of Fear Generalization with fMRI Using Multivariate Pattern Similarity Analysis
Investigation of Fear Generalization with fMRI Using Multivariate Pattern Similarity Analysis
Onat S., Büchel C.
Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf
Generalization of aversive events to previously neutral events relates to problems in anxiety regulation. This is generally investigated by using simplified stimuli that are parametrically modified to span a similarity space with respect to the aversive event and measuring the tuning width of autonomic responses. However on the neuronal level our knowledge about which brain regions underpin this generalization is scarce. Multivariate pattern analysis of BOLD responses offer a valuable tool to investigate brain regions which are responsible in controlling the fear generalization.
We investigated fear generalization using faces as ecologically relevant stimuli using an event-related fMRI design using healthy humans. These stimuli were organized along a circular similarity continuum and two opposing faces were randomly selected as the aversive and neutral face for each participant. Autonomic measures of anxiety such as skin conductance and pupil size showed a gradual decrease in amplitude with increasing distance to the aversive face. These fear tuning profiles were well-captured with a circular Gaussian function that showed a large natural variability in the width of the generalization profiles across participants.
Using multi-variate similarity analysis at the single subject level, we evaluated which brain regions exhibited a similarity gradient with respect to the neuronal activity patterns evoked by the aversive face. In cingulate cortex, putamen, insular cortex and frontal regions, including medial frontal and orbito-frontal cortex, the similarity of neuronal activity patterns decayed with increasing distance from the aversive face and reached lowest values for the neutral face.
Subscribe to:
Posts (Atom)