Annelies Baeck

Distributed subordinate specificity for bodies, faces, and buildings in human ventral visual cortex.

Previous studies have revealed regions in human visual cortex with a strong preference for faces, headless bodies, and buildings. We investigated whether the pattern of activity in these category-selective regions is related to more subordinate distinctions among objects. Our experiments included two types of faces (elderly faces and baby faces), body parts (hands and torsos), and buildings (rural buildings and skyscrapers). Multi-voxel pattern analyses revealed very clear differences in the activation pattern between hands and torsos, and smaller but significant differences in the activation pattern between the two face conditions and between the two building conditions. The subordinate specificity was very distributed, as all category-selective regions were most selective for the distinction between hands and torsos, independently from their preferred category. The selectivity for hands versus torsos was preserved across exemplars and image orientations in all category-selective regions, indicating that the distributed subordinate selectivity is related to relatively invariant and higher-order properties of the images.

Transfer of object learning across distinct visual learning paradigms

Perception of visual stimuli improves with experience. For objects, learning is specific for the stimuli used during training. This is shown in perceptual learning paradigms in which visual perception is challenged by degrading the stimuli, e.g. by backward masking or adding simultaneous noise. We present the first study designed to investigate whether visual object learning is specific to the type of stimulus degradation used. Sixteen participants were trained to recognize common objects. Half of them was trained in a backward masking paradigm, the other half in a simultaneous noise addition paradigm. After five days, performance was measured in four tests: (1) the trained paradigm with the trained objects, (2) the trained paradigm with new objects, (3) the untrained paradigm with the trained objects and (4) the untrained paradigm with new objects. Training effects were specific for the trained objects. In addition, an object-specific transfer to the untrained paradigm was found. The group trained in the simultaneous noise addition paradigm showed a complete transfer of performance to the backward masking task. The transfer was only partial when reversed. This pattern of results indicates that both general processes and processes specific for the type of stimulus degradation are involved in perceptual learning.

The distributed representation of random and meaningful object pairs in human occipitotemporal cortex: The weighted average as a general rule

Natural scenes typically contain multiple visual objects, often in interaction, such as when a bottle is used to fill a glass. Previous studies disagree about the representation of multiple objects and the role of object position herein, nor did they pinpoint the effect of potential interactions between the objects. In an fMRI study, we presented four single objects in two different positions and object pairs consisting of all possible combinations of the single objects. Objects pairs could form either a meaningful action configuration in which they interact with each other or a non-meaningful configuration. We found that for single objects and object pairs both identity and position were represented in multi-voxel activity patterns in LOC. The response patterns of object pairs were best predicted by a weighted average of the response patterns of the constituent objects, with the strongest single-object response (the max response) weighted more than the min response. The difference in weight between the max and the min object was larger for familiar action pairs than for other pairs when participants attended to the configuration. A weighted average thus relates the response patterns of object pairs to the response patterns of single objects, even when the objects interact.

Brain-decoding fMRI reveals how wholes relate to the sum of parts

The human brain performs many nonlinear operations in order to extract relevant information from local inputs. How can we observe and quantify these effects within and across large patches of cortex? In this paper, we discuss the application of multi-voxel pattern analysis (MVPA) in functional magnetic resonance imaging (fMRI) to address this issue. Specifically, we show how MVPA (i) allows to compare various possibilities of part combinations into wholes, such as taking the mean, weighted mean, or the maximum of responses to the parts; (ii) can be used to quantify the parameters of these combinations; and (iii) can be applied in various experimental paradigms. Through these procedures, fMRI helps to obtain a computational understanding of how local information is integrated into larger wholes in various cortical regions.

The transfer of object learning across exemplars and their orientation is related to perceptual similarity

Recognition of objects improves after training. The exact characteristics of this visual learning process remain unclear. We examined to which extent object learning depends on the exact exemplar and orientation used during training. Participants were trained to name object pictures at as short a picture presentation time as possible. The required presentation time diminished over training. After training participants were tested with a completely new set of objects as well as with two variants of the trained object set, namely an orientation change and a change of the exact exemplar shown. Both manipulations led to a decrease in performance compared to the original picture set. Nevertheless, performance with the manipulated versions of the trained stimuli was better than performance with the completely new set, at least when only one manipulation was performed. Amount of transfer to new images of an object was related to perceptual similarity, but not to pixel overlap or to measurements of similarity in the different layers of a popular hierarchical object recognition model (HMAX). Thus, object learning generalizes only partially over changes in exemplars and orientation, which is consistent with the tuning properties of neurons in object-selective cortical regions and the role of perceptual similarity in these representations.

The Transfer of Object Learning after Training with Multiple Exemplars

Object recognition improves with training. This training effect only partially generalizes to untrained images of the trained objects (new exemplars, orientation,…). The aim of this study is to investigate whether and to what extent the learning transfer improves when participants are trained with more exemplars of an object. Participants were trained to recognize two sets of stimuli using a backward masking paradigm. During training with the first set, only one exemplar of each object was presented. The second set was trained using four exemplars of each object. After 3 days of training, participants were tested on all the trained exemplars and a completely new exemplar of the same objects. In addition, recognition performance was compared to a set of completely new objects. For the objects of which four exemplars were used during training, participants showed more generalization toward new exemplars compared to when they were only trained with one exemplar. Part of the generalization effect extended to completely new objects. In conclusion, more variation during training leads to more generalization toward new visual stimuli.

From What Scale of Representation Does Multivariate Pattern Analysis Decode Information

Functional magnetic resonance imaging (fMRI) is severely limited in the extent of detail that it can reveal. In practice, most fMRI studies investigate large areas subtending centimeters, such as the primary visual cortex or face-selective regions. Electrophysiological recordings, however, have