Perceptual Learning of Contrast Detection in the Human Lateral Geniculate Nucleus

Abstract

The brain is continuously modified by perceptual experience throughout life. Perceptual learning, which refers to the long-term performance improvement resulting from practice, has been widely used as a paradigm to study experience-dependent brain plasticity in adults [1, 2]. In the visual system, adult plasticity is largely believed to be restricted to the cortex, with subcortical structures losing their capacity for change after a critical period of development [3, 4]. Although various cortical mechanisms have been shown to mediate visual perceptual learning [5-12], there has been no reported investigation of perceptual learning in subcortical nuclei. Here, human subjects were trained on a contrast detection task for 30 days, leading to a significant contrast sensitivity improvement that was specific to the trained eye and the trained visual hemifield. Training also resulted in an eye- and hemifield-specific fMRI signal increase to low-contrast patterns in the magnocellular layers of the lateral geniculate nucleus (LGN), even when subjects did not pay attention to the patterns. Such an increase was absent in the parvocellular layers of the LGN and visual cortical areas. Furthermore, the behavioral benefit significantly correlated with the neural enhancement. These findings suggest that LGN signals can be amplified by training to detect faint patterns. Neural plasticity induced by perceptual learning in human adults might not be confined to the cortical level but might occur as early as at the thalamic level.

Deep Neural Networks for Modeling Visual Perceptual Learning

“…Post hoc analysis showed a significant increase in weight change proportions in layers 4 and 5 and a significant decrease in layer 2 (Mann-Whitney U, threshold p ϭ 0.01, Bonferroni-corrected for 5 or 4 layers). Therefore, more weight change happened in lower layers when learning the "low-level" Gabor task, and higher layers acquired more change in the "high-level" face task, consistent with theories of VPL (Ahissar andHochstein, 1997, 2004;Watanabe and Sasaki, 2015).…”

“…In addition, DNNs may provide a straightforward way to model the "when" and "where" aspects of VPL that would be otherwise difficult to test experimentally on subjects. As discussed in previous reviews (Watanabe and Sasaki, 2015;, it is likely that VPL involves more areas than the twoor three-layer models of pattern-matching representation and nonlinear readout that typify the field (Dosher and Lu, 2017). The distribution and time course of plasticity could be further examined in other tasks using other DNNs or layer change measures.…”

Layer-dependent multiplicative effects of spatial attention on contrast responses in human early visual cortex

“…This finding suggests a distinct mechanism of attention associated with feedforward visual activity through subcortical neural circuits (O'Connor, Fukui et al 2002, McAlonan, Cavanaugh et al 2008. Attention increased contrast sensitivity in V1 middle layer is also consistent with a previous study showing that visual perceptual learning increased the contrast sensitivity of Magnocellular response of the human LGN (Yu, Zhang et al 2016). A recent monkey electrophysiology study also found different neural mechanisms of attention modulation in the superficial and middle layers of V4 (Nandy, Nassi et al 2017).…”

Patching and Suppression in Amblyopia: One Mechanism or Two?

“…There has been evidence that visual PL could improve an amblyopic eye's Vernier acuity (Levi et al, 1997), visual acuity (Polat et al, 2004), contrast sensitivity (Zhou et al, 2006), and other vision functions (Li and Levi, 2004;Levi, 2005;Huang et al, 2008;Astle et al, 2011). There is evidence that PL can enhance modulation in neuronal tuning in V1 to the trained stimulus and that this consequently facilitates visual functions (Huang et al, 2008;Zhou et al, 2012;Ren et al, 2016;Yu et al, 2016). The role of interocular suppression is less clear for this approach.…”