The theory of two visual systems was firstly proposed in 1960s, to explain the distinct
neural mechanisms underlying visual discrimination and localization in rodents and
frogs. A recent study demonstrated that fear learning could be transmitted either
through visual cortex or simply superior colliculus to the amygdala.
Early brain lesion studies found that tectal lesion is associated with the dysfunction
of object localization and the visual cortical lesion is associated with failure in
pattern discrimination (Schneider, 1969; Ingle, 1973). Similar idea was proposed for
human visual system as well: the two streams hypothesis (Botez, 1975; Milner and Goodale,
2008). It is believed that the dorsal visual system receives whole retinal inputs
in fast transmission manner, contributing to visually guided behaviors; while the
ventral visual system creates imagery with spatial details, and is highly relevant
to visual consciousness.
The traditional view suggested that fear cues are evaluated by visual cortex and then
transmitted to the amygdala, through the visual thalamus. However, it is then realized
that healthy human subjects could detect “unseen” fearful cues through subcortical
connections between right amygdala, pulvinar, and superior colliculus (SC) (Morris
et al., 1999; Tamietto and de Gelder, 2010). This highlighted the possibility of “non-conscious”
or cortex-independent processing of visual fear cues. The hypothesis is further proved
on human patients with cortical blindness, which showed intact fear learning to the
visual cue (Hamm et al., 2003). These evidences argued for the presence of subcortical
pathway in visual cue dependent- fear learning. Lesions of lateral geniculate body
(LG) and lateral posterior nucleus (LP, pulvinar like structure in rodent) of the
thalamus together, but not respectively, led to impairment of visual cue-paired fear
conditioning (Shi and Davis, 2001). Yet, the exact neural circuits underlying the
visual fear guided behavior was unknown. A recent study employed optogenetic dissection
of such circuits to answer the question.
One previous study investigated the innate defensive response of mice to overhead
looming stimuli as the fear behavior paradigm (Wei et al., 2015). When the authors
silenced a group of excitatory neurons by expression and following light activation
of NpHR in the superficial layer of SC, the innate fear behavior is eliminated. The
authors then injected retrograde trans-synaptic tracers into the amygdala, and identified
LPN as the relay in SC-amygdala pathway. Indeed, activating SC-expressed ChR2 axon
terminals in LPN was sufficient to elicit the freezing behavior. Collectively, their
results suggested that the excitatory pathway from SC to LPN, and then the LPN to
amygdala pathway is critical for the amygdala expression of this visual fear cue guided
behavior. It is still unknown if the LPN-amygdala pathway is excitatory or inhibitory.
Interestingly, Wei et al. found that Parvalbumin (PV) positive neurons (classical
cortical fast-spiking interneurons) in SC are also transfected by CaMKII-promoter
virus. In the recent study, Shang et al. further examined these PV neurons and found
that SC PV neurons have distinct properties to cortical PV neurons in terms of morphology,
electrophysiological characteristics, and biochemical content (Shang et al., 2015).
These SC PV neurons are found to be excitatory, and the optogenetic activation of
these neurons raised fear response behaviors. Anatomical tracing study confirmed that
these SC PV neurons project to parabigeminal nucleus (PBGN), which then project to
amygdala. Optogenetic activation of axon terminals from SC PV neuron within PBGN could
generate the fear response, confirming that the PV+ SC-PBGN-Amygdala is a non-canonical
fear pathway to visual cues. It is suggested that the expression of Parvalbumin is
critical to drive the fast response required for this pathway (Figure 1).
Fear learning through the two visual systems. The visual discrimination system receives
information from visual thalamus and transmits the information to visual cortex for
fear detection, before reaching amygdala. On the other hand, the superior colliculus
could directly excite amygdala neurons through LPN or BPGN pathway, potentially in
an “unconscious” manner.
The results of Shang et al. may speak to the long debate on the visual mechanism underlying
fear conditioning. In particular, is visual cortex required for the fear response
behavior to visual cues? The brain lesion study or physiological recordings showing
that SC neurons respond to fear behavior could be explained by compensation mechanism
or reciprocal neural circuits between visual cortex and the SC or SC-projecting brain
regions. For a stronger test of the hypothesis, future studies may wish to combine
optogenetic excitation of SC pathway and optogenetic silencing of the visual cortex
and LGN pathway, or with muscimol infusion into the visual cortex to exclude the collateral
excitation or reciprocal neural circuits involved in fear responses described above.
Knowing that the fear learning can be mediated by the SC-LPN/PBGN-Amygdala pathway
raises a new set of questions. How does the pathway discriminate fearful or neutral
information? Are there any special retina ganglion cells responsible for the fear
response behavior? What other pathways may brain employ in parallel? For instance,
direct retinal projection to dorsal raphe is known to regulate the affective behavior
(Ren et al., 2013). We now have the optogenetic tools to answer these possibilities,
and to answer the psychological theory “two visual systems” from the fear learning
Conflict of interest statement
The authors declare that the research was conducted in the absence of any commercial
or financial relationships that could be construed as a potential conflict of interest.