1. Attentional Modulation of Neuronal Activity Depends on Neuronal Feature Selectivity 
Jacqueline R. Hembrook-Short, Vanessa L. Mock, Farran Briggs 
Current Biology 
Volume 27, Issue 13, Pages e5 (July 2017)
DOI: /j.cub
Copyright © 2017 Elsevier Ltd Terms and Conditions

2. Figure 1 Attention Task and Behavioral Results
(A) Schematic screen shots of the contrast-change detection task. A red central fixation dot (left) cued the monkey to attend to the drifting sinusoidal grating in the lower hemifield, while a blue fixation dot (right) cued the monkey to attend to the grating in the upper hemifield. Dashed circle (not displayed in the actual task) indicates receptive field location of recorded neurons. The timeline for a trial is indicated far right.
(B) Accuracy, as average percentage correct, for each of the three monkeys on validly cued (attend-toward and attend-away conditions in red and blue, respectively) and invalidly cued (gray) trials. Error bars represent SEMs. Asterisks indicate significant reductions in accuracy on invalidly cued trials (monkey B: n = 95 sessions, p = 1.9 × 10−10; attend-toward = 82% ± 1%, attend-away = 82% ± 1%, invalid = 63% ± 3%; monkey O: n = 23 sessions, p = 0.007, attend-toward = 65% ± 4%, attend-away = 68% ± 5%, invalid = 25% ± 13%; monkey E: n = 40 sessions, p = 0.001, attend-toward = 70% ± 3%, attend-away = 71% ± 3%, invalid = 45% ± 6%).
Current Biology  , e5DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

3. Figure 2
Example Single-Unit Recordings and Attentional Modulation of Neuronal Firing Rate
(A) Schematic representation of the linear array positioned to record from V1 neurons spanning the cortical layers. Average local field potential (LFP) responses to a flashed stimulus in the receptive field (flash onset indicated by vertical black line). Horizontal dashed line highlights the polarity reversal at LFP response onset, indicative of the layer 4C/5 border. See Figure S1 for receptive field locations of recorded neurons.
(B) Current source density spectrum generated from LFPs traces in (A). Vertical black line again indicates flash onset at time zero. Horizontal dashed lines indicate borders of laminar compartments: supragranular (SG), granular (G), and infragranular (IG).
(C) Schematic representation of linear array with contacts assigned to SG (green), G (orange), and IG (purple) laminar compartments. Upper-right images shows two single-unit waveforms (a simple [cyan] and a complex [magenta] neuron) isolated from one SG contact and the principal components analysis clustering of first and second principal components of the same waveforms (both waveforms are significantly different from each other and from the noise [gray cluster] at p = 5.7 × 10−31; waveform SNRs were 9.2 and 4.8). Lower-right image shows the inter-spike-interval (ISI) distributions for the two waveforms and noise. Red dashed lines illustrate short-ISI cutoff, and red bar above noise ISI distribution illustrates a high percentage of short ISI violations (3.6%) compared to 0% short ISI violation for the cyan and magenta waveforms.
(D) Peri-stimulus time histograms (PSTHs) for 11 simultaneously recorded neurons spanning the SG (green), G (orange), and IG (purple) laminar compartments (contact number color-coded by laminar compartment at right) from a single session. Average spike count (in 1-ms bins) prior to and following grating onset at time zero is illustrated separately for attend-toward (red) and attend-away (blue) trials.
(E) Additional example PSTHs for four neurons, three facilitated (top three examples), and one suppressed (bottom example) by attention. PSTHs illustrate spike counts prior to and following grating onset at time zero on attend-toward (red) and attend-away (blue) trials. AI values for each neuron are indicated along with laminar compartment location and simple/complex type. Black curves underneath PSTHs illustrate differential modulation by attention (attend-toward minus attend-away), with attentional modulation during the grating display period illustrated by black fills, using the same scaling as above PSTHs.
(F) Average firing rate for simple neurons (top; n = 74) and complex neurons (bottom; n = 94) just prior to and during grating presentation (grating onset at time zero) on attend-toward (red) and attend-away (blue) trials. Shaded regions represent SEMs.
Current Biology  , e5DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

4. Figure 3
Attentional Modulation of Firing Rate across Neuronal Types and Laminar Locations
(A) Distributions of attention index (AI) values for simple neurons (white/light gray; n = 74) and complex neurons (black/dark gray; n = 94). Gray tones indicate neurons that were significantly modulated by attention (n = 6 complex neurons; n = 11 simple neurons). Average simple neuron AI = −0.004 ± (dashed line), and average complex neuron AI = 0.03 ± (solid line).
(B) AI values were significantly greater for complex neurons compared to simple neurons (asterisk, p = ). Error bars represent SEMs.
(C) Average AI values for simple (open bars) and complex (solid bars) neurons separated by laminar compartment location (supragranular, SG in green; granular, G, in orange; infragranular, IG, in purple). Error bars represent SEMs. Asterisk indicates statistically significant differences in attentional modulation of firing rate across neuronal types and laminar locations (p = 0.0046). Specifically, complex neurons in the G and IG laminar compartments were significantly more modulated by attention compared to simple neurons in the SG and IG laminar compartments. SG simple: n = 26, average AI = −0.008 ± 0.009; SG complex: n = 37, average AI = 0.018 ± 0.007; G simple: n = 28, average AI = 0.005 ± 0.01; G complex: n = 25, average AI = 0.029 ± 0.008; IG simple: n = 20, average AI = −0.012 ± 0.013; IG complex: n = 32, average AI = 0.037 ±
Current Biology  , e5DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

5. Figure 4 Tuning Curves for Six Representative V1 Neurons
Each row illustrates orientation, contrast, spatial frequency, temporal frequency, and size tuning curves for a single V1 neuron with laminar compartment location, simple/complex type, and AI value listed above the middle (spatial frequency) plot. Data are dots, and error bars represent SEMs. Lines are curve fits (Gaussian fits for orientation and size tuning data; power fits for contrast tuning data; and smoothing spline fits for spatial and temporal frequency data).
Current Biology  , e5DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

6. Figure 5
Relationships between Neuronal Feature Selectivity and Attentional Modulation of Firing Rate
(A) Negative relationship, indicated by dashed line illustrating regression fit, between AI and c50, or contrast to evoke a half-maximal response, for V1 neurons (simple, open dots; complex, closed dots; R2 = 0.03, p = 0.04, n = 135).
(B) Positive relationship, indicated by dashed line illustrating regression fit, between AI and direction selectivity index (DSI) for V1 neurons (R2 = 0.02, p = 0.04, n = 168).
(C) Significant interaction between AI, DSI, and c50 illustrated by polynomial fit and three-way regression model (R2 = 0.04, p = 0.016, n = 143) for V1 neurons (black dots).
(D–F) Significant relationships (solid lines illustrate regression fits) were observed between AI and orientation HWHH for SG complex neurons (D, R2 = 0.11, p = 0.04, n = 37), between AI and DSI for G complex neurons (E, R2 = 0.32, p = 0.006, n = 22), and between AI and SSI for G complex neurons (F, R2 = 0.22, p = 0.04, n = 18).
(G–I) Relationships (solid lines illustrate regression fits, R2 ≤ 0.1, p ≥ 0.2 for all) between AI and orientation HWHH for G and IG complex neurons (G), DSI for SG and IG complex neurons (H), and SSI for SG and IG complex neurons (I).
See Figure S2 and Table S1 for all other comparisons. See Table 1 for full regression statistics corresponding to each plot.
Current Biology  , e5DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

7. Figure 6
Attentional Modulation of Variance and Contrast Sensitivity among V1 Neurons
(A) Average Fano factor across attention conditions (red, attend-toward; blue, attend-away) for simple (open bars) and complex (closed bars) neurons located in each laminar compartment (x axis labels). Error bars represent SEMs. No differences in Fano factor were observed across neuronal types, laminar compartments, or attention conditions (p = 0.6; SG simple: n = 26, attend-toward Fano factors = 3.1 ± 0.6, attend-away Fano factors = 3.9 ± 0.8; G simple: n = 28, attend-toward Fano factors = 2.8 ± 0.6, attend-away Fano factors = 2.9 ± 0.6; IG simple: n = 20, attend-toward Fano factors = 2.8 ± 0.4, attend-away Fano factors = 2.7 ± 0.4; SG complex: n = 37, attend-toward Fano factors = 3.5 ± 0.5, attend-away Fano factors = 3.3 ± 0.5; G complex: n = 25, attend-toward Fano factors = 3.8 ± 0.7, attend-away Fano factors = 3.8 ± 0.7; IG complex: n = 32, attend-toward Fano factors = 2.3 ± 0.3, attend-away Fano factors = 2.1 ± 0.3). However, Fano factors were lowest among IG complex neurons.
(B) Average spike count correlations for 1,239 total pairs of simultaneously recorded V1 neurons across attention conditions (p = 0.19; average spike count correlations attend-toward = 0.05 ± 0.008; average spike count correlations attend-away = 0.06 ± 0.008). Error bars represent SEMs.
(C) Average c50 values for simple (open bars) and complex (closed bars) neurons across laminar compartments. Error bars represent SEMs. There were no differences in contrast sensitivity across neuronal types or laminar compartments (p = 0.5; SG simple: n = 26, average c50 = 23% ± 5%; SG complex: n = 35, average c50 = 15% ± 3%; G simple: n = 12, average c50 = 11% ± 2%; G complex: n = 20, average c50 = 12% ± 3%; IG simple: n = 17, average c50 = 10% ± 1%; IG complex: n = 31, average c50 = 18% ± 4%).
(D) Significant positive relationship (black line illustrates regression fit; R2 = 0.2, p = 0.007) between neuronal contrast discriminability (ROC AUC) and AI for 33 simple (open dots) and complex (closed dots) neurons.
Current Biology  , e5DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions