Abstract
The early visual system is composed of spatial frequency-tuned channels that break an image into its individual frequency components. Therefore, researchers commonly filter images for spatial frequencies to arrive at conclusions about the differential importance of high versus and low spatial frequency image content. Here, we show how simple decisions about the filtering of the images, and how they are displayed on the screen, can result in drastically different behavioral outcomes. We show that jointly normalizing the contrast of the stimuli is critical in order to draw accurate conclusions about the influence of the different spatial frequencies, as images of the real world naturally have higher contrast energy at low than high spatial frequencies. Furthermore, the specific choice of filter shape can result in contradictory results about whether high or low spatial frequencies are more useful for understanding image content. Finally, we show that the manner in which the high spatial frequency content is displayed on the screen influences how recognizable an image is. Previous findings that make claims about the visual system’s use of certain spatial frequency bands should be revisited, especially if their methods sections do not make clear what filtering choices were made.
Highlights
Since the discovery of distinct spatial frequency channels in the human visual cortex [1], select spatial frequencies have been suspected to be involved in a variety of perceptual phenomena, including the processing of scene gist versus detailed scene information [2,3], in the processing of emotions in images of faces [4,5,6,7], as well as processing of objects in scene context [8]
Planned paired t tests show that the difference in accuracy between high spatial frequencies (HSF) and Low spatial frequencies (LSF) images is highly significant without contrast normalization (t(20) = −13.11; p = 2.79 × 10−11 ) but not when normalizing the filtered images for RMS contrast (t(20) = 0.90; p = 0.378)
The direction of the difference switches from the Heaviside filter (HSF < LSF) to the Gaussian filter (HSF > LSF). The latter result is presumably due the spillover of LSF image content into the Gaussian HSF image (Figure 4B, top right). These results powerfully demonstrate that the particular choice of the frequency filter can pre-determine the result of comparisons between high and low spatial frequencies
Summary
Since the discovery of distinct spatial frequency channels in the human visual cortex [1], select spatial frequencies have been suspected to be involved in a variety of perceptual phenomena, including the processing of scene gist versus detailed scene information [2,3], in the processing of emotions in images of faces [4,5,6,7], as well as processing of objects in scene context [8]. Low and high spatial frequencies carry different aspects of the information contained in an image. Low spatial frequencies (LSF) concern global shape, whereas high spatial frequencies (HSF) carry edges, contours, and more detailed aspects. Evidence for the role of LSF versus HSF in scene perception from functional magnetic resonance imaging (fMRI) research is mixed. Several studies have shown higher activation of the scene-selective parahippocampal place area (PPA) for LSF than HSF [9,10], whereas other studies have
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
