Bernal S D_2010.pdf - University of Plymouth

2.2. HIGH-LEVEL FEEDBACK
Thirdly, physiological findings, some of which are described further on in this section, indicate
feudback plays a modulatory rather than a driving role.
A number of clarifications and remarks on Macknik's theory are now described. Firstly, ii is
impohani lo make clear that the feedforward projection from an unoriented LtiN cell to several
VI oriented neurons consists of a single axonal fibre which branches at the end to make the
different synaptic connections. However, each VI neuron requires an individual axona! fibre to
feedback to the original LGN neuron. Thus, although the number of feedforward and feedback
synaptic connections is etjuivaienl, a larger number of feedback axonal (ibrcs is required. The
explanatory diagram included in figure 81.3 in Macknik and Martinez-Conde (2009) shows a
one-to-one relationship between feedforward and feedback connections (depicted as arrows),
which contrasts with the one-to-many relationship described in the text, and may lead to confu­
sion.
Furthermore, it is not clear whether Macknik's principle generalizes to higher extrastriate areas,
presumably with a larger feature space at each location and consequently a more sparsely dis­
tributed connectivity pattern. For instance, it seems unlikely that a neuron coding for a specific
orientation in VI receives feedback from all the complex features coded in V4 at that location.
Overall. Macknik's claims seem reasonable and provide an explanation for the large ratio of
feedback to feedforward projections in the visual system. Nonetheless, after taking into ac­
count this consideration, the effective connectivity ratio is still significant (one-to-one), thus
still constituting an argument for a potentially .strong functional role for feedback connections.
With respect to the relatively weak modulatory effects attributed to feedback, it must be noted
that the modulatory strength might be dependent on the characteristics and context of the input
stimuli. For example, the weight of top-down feedback might be stronger for more ambiguous
images,
2.2.1.2 ]*hysiolo|;icijl evidence
In line with the functional role of feedback suggested by anatomical considerations, from a
physiological perspective, there exists abundant evidence showing the modulatory effects of
high-level feedback on lower levels. The most simple example comes from cortical areas feed-
26