Luo and Maunsell used a specially designed task to show that the only effect of attention in visual cortex is to improve the signal-to-noise ratio of neurons.

Neuronal Modulations in Visual Cortex Are Associated with Only One of Multiple Components of Attention
TZ Luo, JHR Maunsell – Neuron, 2015

See highlight by Timothy J. Buschman:
Paying Attention to the Details of Attention
TJ Buschman – Neuron, 2015

Our lab and others (e.g., Buschman and Miller, 2007; Bastos et al 2012) has suggested that top-down (feedback) vs bottom-up (feedforward) cortical processing is mediated by synchrony between cortical areas at different frequencies: lower (e.g., beta band) for top-down vs higher (e.g., gamma band) for bottom-up.  These two different frequency bands allow top-down vs bottom-signals to multiplex through the same circuits, much as different FM radio stations multiplex through the airwaves.  They may also allow cortical microcircuits to engage in helpful things like predictive coding (Bastos et al., 2012).

Schmiedt et al (2014) provide new evidence for this.    They recorded neural activity in visual area V4 after damage to primary visual area V1.  V4 is higher in the cortical hierarchy, so V1 has a bottom-up influence on V4.  They found that damage to V1 decreased the gamma in V4 that follows appearance of a visual stimulus.  That is consistent with gamma carrying bottom-up or feedforward signals, lost after V1 damage.  By contrast, V4 beta activity was minimally affected, reflecting the unaffected top-down influence on V4   Normally there is beta suppression during visual stimulation, presumably because the bottom-up inputs overwhelm or suppress beta-mediated top-down processing.  After V1 damage, this suppression of top-down beta rhythms was diminished, presumably because it was no longer suppressed by bottom-up influences from V1.

For further reading:
Buschman, T.J. and Miller, E.K. (2007) Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices. Science. 315: 1860-1862  The Scientist’s “Hot Paper” for October 2009. View PDF »

Bastos AM, Usrey WM, Adams RA, Mangun GR, Fries P, Friston KJ. Canonical microcircuits for predictive coding. Neuron. 2012 Nov 21;76(4):695-711. doi:
10.1016/j.neuron.2012.10.038. Review.