Elsevier

NeuroImage

Volume 126, 1 February 2016, Pages 173-183
NeuroImage

Spatial properties of objects predict patterns of neural response in the ventral visual pathway

https://doi.org/10.1016/j.neuroimage.2015.11.043Get rights and content

Highlights

  • Distinct neural patterns were evident to different object categories in ventral visual stream.

  • Changing the size of the images had a significant effect on these patterns.

  • A computational analysis shows that spatial properties of images can explain patterns of neural response to different objects.

  • The effect of spatial properties declined from posterior to anterior regions of the ventral stream.

Abstract

Neuroimaging studies have revealed topographically organised patterns of response to different objects in the ventral visual pathway. These patterns are thought to be based on the form of the object. However, it is not clear what dimensions of object form are important. Here, we determined the extent to which spatial properties (energy across the image) could explain patterns of response in these regions. We compared patterns of fMRI response to images from different object categories presented at different retinal sizes. Although distinct neural patterns were evident to different object categories, changing the size (and thus the spatial properties) of the images had a significant effect on these patterns. Next, we used a computational approach to determine whether more fine-grained differences in the spatial properties can explain the patterns of neural response to different objects. We found that the spatial properties of the image were able to predict patterns of neural response, even when categorical factors were removed from the analysis. We also found that the effect of spatial properties on the patterns of response varies across the ventral visual pathway. These results show how spatial properties can be an important organising principle in the topography of the ventral visual pathway.

Introduction

Although a large number of studies have investigated the functional properties of high-level regions in the ventral visual pathway, an overarching framework for how information is represented topographically in this region is not fully resolved (Op de Beeck et al., 2008, Grill-Spector and Weiner, 2014). Neuroimaging studies using univariate analyses have shown that regions of the ventral visual pathway are selective for different categories of objects (Kanwisher, 2010). The location of these regions is broadly consistent across individuals, suggesting that common organising principles underpin the topography of this region (Malach et al., 1995, Kanwisher et al., 1997, Epstein and Kanwisher, 1998, Cohen et al., 2000, Downing et al., 2001). The importance of this topography is evident in multivariate studies that have shown that the pattern of response across the entire ventral stream can distinguish a greater range of object categories than shown in previous univariate studies (Haxby et al., 2001, Spiridon and Kanwisher, 2002, Kriegeskorte et al., 2008b). The potential importance of the pattern of response is demonstrated by the fact that the ability to discriminate particular object categories is still evident when the most category-selective regions are removed from the analysis (Haxby et al., 2001).

The distributed nature of the fMRI response to different categories of objects within the ventral visual pathway has been interpreted as showing a topographic map of object forms; a hypothesis known as ‘object form topography’ (Haxby et al., 2001). However, it is not clear what dimensions are important for this object form topography. A variety of evidence has suggested that patterns of response in the ventral visual pathway are linked to the categorical or semantic information that the images convey (Kriegeskorte et al., 2008b, Naselaris et al., 2009, Connolly et al., 2012). Evidence for other organising principles can be found in the large-scale patterns of response to animacy (Chao et al., 1999, Kriegeskorte et al., 2008b, Clarke and Tyler, 2014) or to the real-world size of objects (Konkle and Oliva, 2012, Konkle and Caramazza, 2013). However, it remains unclear how these higher level categorical or semantic properties in the ventral visual pathway might arise from the image-based representations found in early visual regions (Op de Beeck et al., 2008).

In recent studies, we have asked to what extent image-based representations might underpin category-selective patterns of response in the ventral visual pathway (Rice et al., 2014, Watson et al., 2014). We found that similarities in the patterns of fMRI response between different categories of objects could be predicted by corresponding similarities in their low-level image properties. However, more needs to be known about which image properties are important. Previous studies have suggested that ventral temporal cortex may be sensitive to the spatial properties of the visual scene (Levy et al., 2001, Malach et al., 2002, Golomb and Kanwisher, 2012, Cichy et al., 2013, Troiani et al., 2014), but it has not been clear whether this reflects a modification of the underlying categorical organisation based on the way natural object categories are viewed or whether spatial properties themselves represent a fundamental organising principle in the ventral visual pathway (Kanwisher, 2001).

The aim of this study was to systematically investigate the importance of spatial properties to the pattern of response across the ventral visual pathway. To address this question, we determined the effect of changes in the actual size of the image on the patterns of response to different types of object using multi-voxel pattern analysis (MVPA). The image size manipulation is known to have a negligible effect on the perception of objects but will clearly have a profound effect on the spatial energy across the image. If the pattern of response is invariant to changes in image size then we would expect similar patterns of response to the same object category presented at different image sizes. If, on the other hand, the patterns of response are sensitive to changes in image size, we would expect significant differences in the pattern of response to the same object presented at different image sizes. Next, we used a computational approach to more directly measure energy across the image. Our aim was to determine whether more subtle differences in spatial properties beyond those carried by the large changes conveyed by size could explain patterns of response in the ventral visual pathway.

Section snippets

Participants

20 participants (8 males; median age 26; age range 19–36) took part in the experiment. All participants were neurologically healthy, right-handed, and had normal or corrected-to-normal vision. Written consent was obtained for all participants and the study was approved by the York Neuroimaging Centre Ethics Committee.

Stimuli

Visual stimuli were back-projected onto a custom in-bore acrylic screen at a distance of approximately 57 cm from the participant with all images subtending approximately 10.7° of

Results

Patterns of neural response were measured for 4 different object categories (bottle, chair, face, house) presented at 2 different sizes (large, small). Behavioural performance on the one-back task was near the ceiling (mean accuracy = 97.40%, SEM = .41%), indicating that participants maintained attention throughout the scan. Fig. 4a shows the normalised group responses to each condition across the ventro-temporal ROI. Responses above the mean are shown in red, and responses below the mean are shown

Discussion

The aim of this study was to investigate how spatial properties of images influence patterns of neural response in the ventral visual pathway. We found that patterns of neural response to images of different object categories are influenced by the actual size of the images. We then used a computational approach to more directly test whether more subtle variance in the spatial properties of images could explain the different patterns of response to objects in the ventral temporal cortex. Our

Acknowledgments

We would like to thank Danai Beintari for her assistance with data collection and analysis, and Grace Rice for her work in developing the stimulus set.

References (46)

  • M. Spiridon et al.

    How distributed is visual category information in human occipito–temporal cortex? An fMRI study

    Neuron

    (2002)
  • B.A. Wandell et al.

    Visual field maps in human cortex

    Neuron

    (2007)
  • D.M. Watson et al.

    Patterns of response to visual scenes are linked to the low-level properties of the image

    Neuroimage

    (2014)
  • T. Bonhoeffer et al.

    Iso-orientation domains in cat visual cortex are arranged in pinwheel-like patterns

    Nature

    (1991)
  • L.L. Chao et al.

    Attribute-based neural substrates in temporal cortex for perceiving and knowing about objects

    Nat. Neurosci.

    (1999)
  • R.M. Cichy et al.

    Probing principles of large-scale object representation: category preference and location encoding

    Hum. Brain Mapp.

    (2013)
  • A. Clarke et al.

    Object-specific semantic coding in human perirhinal cortex

    J. Neurosci.

    (2014)
  • L. Cohen et al.

    The visual word form area

    Brain

    (2000)
  • A.C. Connolly et al.

    The representation of biological classes in the human brain

    J. Neurosci.

    (2012)
  • P.E. Downing et al.

    A cortical area selective for visual processing of the human body

    Science

    (2001)
  • D.M. Drucker et al.

    Different spatial scales of shape similarity representation in lateral and ventral LOC

    Cereb. Cortex

    (2009)
  • R. Epstein et al.

    A cortical representation of the local visual environment

    Nature

    (1998)
  • J.D. Golomb et al.

    Higher level visual cortex represents retinotopic, not spatiotopic, object location

    Cereb. Cortex

    (2012)
  • Cited by (20)

    • Changes in face category induce stronger duration distortion in the temporal oddball paradigm

      2022, Vision Research
      Citation Excerpt :

      This means that the change in face category from the preceding stimuli had an impact on the oddball effect, suggesting that category processing plays a roll in the oddball effect. However, images from different categories also have different low-level visual properties (Andrews, Watson, Rice, & Hartley 2015; Watson, Young, & Andrew, 2016), which can lead to duration distortion (Aaen-Stockdale et al., 2011; Benton & Redfern, 2016; Matthews et al., 2011). To test whether the results of Experiment 1 can be attributed to low-level visual differences between the faces in different categories, Experiment 2 used Fourier phase-scrambled images where configural information is eliminated while the global power of the two-dimensional frequency component is preserved.

    • The cortical face network of the prosopagnosic patient PS with fast periodic stimulation in fMRI

      2019, Cortex
      Citation Excerpt :

      This test will evaluate whether face-selective activation in PS's brain can be attributed even partly to low-level visual information in the stimuli. Such control has become important in light of relatively recent claims that category-selective responses in high-level regions obtained in fMRI with classical face and object localizers could be partly, or mainly, due to low-level visual cues preserved by phase-scrambling of the stimuli (i.e., amplitude spectrum; Andrews, Watson, Rice, & Hartley, 2015; Watson, Young, & Andrews, 2016). Third, we take advantage of the high sensitivity (i.e., SNR) of the novel FPS-fMRI approach and the objectivity of identification of the response of interest (i.e., at a predetermined frequency) to quantify and define the spatial extent of the right lateral fusiform gyrus selective response to faces with respect to the patient's posterior lesion in the right inferior occipital gyrus.

    • Patterns of neural response in face regions are predicted by low-level image properties

      2018, Cortex
      Citation Excerpt :

      The strong linear relationship we observed between low-level image properties and the spatial patterns of response suggests instead that the topographic organization of face regions can also be in large part explained by low-level properties of the image. Our results are consistent with a number of recent studies that have shown that lower-level image properties can be used to predict patterns of response to different object categories in the ventral visual pathway (Andrews et al., 2015; Bracci & de Beeck, 2016; Coggan et al., 2016; O'Toole, Jiang, Abdi, & Haxby, 2005; Rice et al., 2014; Watson et al., 2014, 2016, 2017). For example, the spatial pattern of response of different object categories can be predicted by the image properties of exemplars from that category (Rice et al., 2014; Watson et al., 2014, 2016, 2017).

    View all citing articles on Scopus
    View full text