Animal welfare
The scientists in this study are aware and are committed to the great responsibility they have in ensuring the best possible science with the least possible harm to any animals used in scientific research [38]. Details about our animal care and handling have been reported previously [25, 51]. We summarize relevant details here: The animals were group-housed with other macaque monkeys in facilities of the German Primate Center in Goettingen, Germany, in accordance with all applicable German and European regulations. The facility provides the animals with an enriched environment including a multitude of toys and wooden structures [2, 6], natural as well as artificial light, exceeding the size requirements of the European regulations, including access to outdoor space. Surgeries were performed aseptically under balanced anesthesia using standard techniques, including appropriate peri-surgical analgesia and monitoring to minimize potential suffering. The German Primate Center has several staff veterinarians that regularly monitor and examine the animals and consult on procedures. During the study, the animals had unrestricted access to food and fluid, except on the days where data were collected or the animals were trained on the behavioral paradigm. On these days, the animals were allowed unlimited access to fluid through their performance in the behavioral paradigm. Here, the animals received fluid rewards for every correctly performed trial. Throughout the study, the animals’ psychological and veterinary welfare was monitored by the veterinarians, the animal facility staff, and the lab’s scientists, all specialized in working with non-human primates.
We have established a comprehensive set of measures to ensure that the severity of our experimental procedures falls into the category of mild to moderate, according to the severity categorization of Annex VIII of the European Union’s directive 2010/63/EU on the protection of animals used for scientific purposes (see also [34]).
Monkey surgery
Two male macaque monkeys (Macaca mulatta) were implanted with custom-made titanium head holders to prevent head movement during the recording and gaze tracking. Additionally, they were implanted with a recording chamber (Crist Instruments, Hagerstown, MD, USA, or 3DI, Jena, Germany), first over one and subsequently over the other hemisphere. The position of the recording chambers was planned based on anatomical fMRI scans using a MATLAB-based (The MathWorks, Inc., Natick, MA) software [31].
Apparatus
During recordings, monkeys were seated in a custom-made primate chair in a dark cabin at a distance of 57 cm from the computer monitor (Quato Display 240 m). Visual stimuli were presented with a refresh rate of 60 Hz and a resolution of 1920 × 1200 pixel. Eye position was recorded using an eyetracker (ET49, Thomas Recording, Giessen, Germany) with an acquisition rate of 230 Hz. Acute recording/injection was performed using a multielectrode manipulator equipped with a pressure injection system and containing two recording electrodes and an injection micropipette (MiniMatrix, Thomas Recording). The majority of recordings had inter-tip distance of electrodes and micropipette of approx. 300 μm. Later recordings were performed using an adaptor to achieve a closer spacing of approx. 100 μm, bringing us closer to the spacing used by Herrero et al. [20], which was in the order of 10s of μm. In detail, we used this adaptor in one of the monkeys for both substances, acetylcholine and scopolamine, in 15% and 30% of the cells respectively. Additional file 1: T1 provides a detailed list of the distances used for every recorded neuron. For further details about preparation, handling, and care of the system, see Veith et al. [47]. Recording of neuronal signals and real-time spike sorting was performed with a data acquisition system (MAP, Plexon Inc., Dallas, USA). All stimuli were generated and presented using custom-made software built for real-time visual experiments running on an Apple Macintosh PowerPC. In addition, the software monitored eye position, controlled fluid reward release, and collected behavioral as well as electrophysiological data.
Stimuli
The stimuli used were random dot patterns (RDPs) presented on a uniform gray background (luminance 27 cd/m2). Each RDP consisted of small bright dots (size 0.1 deg, luminance 38 cd/m2, Weber luminance contrast 40%), coherently moving linearly within a circular, stationary aperture. One RDP was placed at the most responsive part of the receptive field (RF), and the second RDP was placed diametrically opposite, relative to the centrally presented fixation point. The speed of the moving dots was adjusted to the recorded cell’s preference within a range of 4–12 deg/s, and the preferred movement direction of the recorded cell was defined based on a pre-experiment tuning session. The radius of the RDP was increased when stimuli had to be placed very eccentrically in order for the stimulus change (see below) to remain detectable for the monkey (range of 2–3 deg).
Substances
In order to manipulate the cholinergic system, we used the general agonist acetylcholine and the muscarinic antagonist scopolamine in various concentrations and volumes (Sigma-Aldrich, St. Louis, MO, USA). All substances were diluted in sterile 0.9% (0.154 mol/l) saline (NaCl, BBraun, Melsungen, Germany). For details, see Veith et al. [47].
For acetylcholine, concentrations of 0.1, 0.15, and 0.2 mol/l and injection rates of 2, 3, and 4 nl/min were used. Scopolamine was injected with concentrations of 0.01, 0.05, and 0.1 mol/l and rates of 1, 2, and 4 nl/min. As a control, saline was injected with a volume of 2, 4, and 6 nl/min in order to mimic the influence of different volumes onto the recorded neuron. Additional file 1: T1 provides a detailed list of the concentration and total injected volume combinations used.
Measured pH values stayed at approx. 5 for all substances used. Different concentrations of the substances used had only weak influences on pH values.
Behavioral task
This task was designed to guide the monkeys’ sustained selective spatial attention to various locations on the monitor. Therefore, the two monkeys were trained to detect a motion direction change in the RDP that was cued at the beginning of the trial. The cue was either placed within the neuron’s receptive field (attend-in) or in the other hemifield (attend-out). In the control (sensory) condition, both RDPs were task irrelevant (attend-fix).
Monkeys initiated each trial by holding a lever and fixating the centrally presented fixation point. In the attend-in and attend-out conditions, the centrally presented fixation point remained red (square with side length 0.16 deg; luminance 14 cd/m2) during the entire trial. After a delay of 150 ms, a static dot pattern, serving as an exogenous cue, was presented for 150 ms at the future stimulus position either within the neuron’s receptive field (attend-in) or outside of it (attend-out). An inter-stimulus interval of 350 ms followed, where only the fixation point was shown. Subsequently, two RDPs were presented on the screen, one placed within the neuron’s receptive field and the other in the opposite hemifield. Both moved linearly in the same direction, either the preferred direction of the recorded neuron or its null direction (preferred direction + 180 deg). Monkeys had to respond to a slight direction change (duration 130 ms) at the cued location (target) and had to ignore a direction change at the uncued location (distractor). The angle of direction change varied from 25 to 35 deg, depending on the stimulus eccentricity, and remained the same within a recording session. The aim was to adjust the task difficulty depending on stimulus position, choosing bigger angles for more eccentric stimuli. The direction change could occur in a time window of 200–2500 ms after stimulus onset, to ensure that spatial attention was maintained at the cued location. In 1/10th of the trials, no direction change happened at the target position and monkeys were rewarded for not responding until the trial ended.
In the sensory condition (attend-fix), the monkeys had to respond to a slight luminance change of the fixation point and had to ignore direction changes of the two shown RDPs. Here, the fixation point changed from red to gray immediately after the trial began. The behaviorally relevant luminance change (from 85 to 52 cd/m2) of the fixation point could occur in a time window of 200–2500 ms after the fixation point color change. Additionally, we had a baseline condition, where only the fixation point was shown on the screen and the monkeys again had to detect a luminance change.
In all conditions, monkeys had to respond by releasing a lever within a fixed time window after target onset and were rewarded with a drop of juice. Trials of all attentional and sensory conditions were presented in random order. A trial was aborted if eye fixation was interrupted or eye gaze moved outside of the fixation window (1.2 deg radius around the fixation spot). Aborted trials were repeated later to allow comparable number of trials per condition.
Experimental procedure
Single unit activity was recorded using two single tungsten electrodes of two different impedances (0.2–0.5 MΩ and 1–2 MΩ) placed in a multielectrode recording system (Thomas Recording, Giessen, Germany). Data was filtered (150 Hz–5 kHz) and amplified (gain range 1000–32,000). Isolation of single units was performed using online window discrimination (RASPUTIN, Plexon Inc., Dallas, TX, USA), and later offline sorted (Plexon Offline Sorter version 3.3.5).
Characterization of isolated cells
In order to confirm isolated cells were in area MT, we examined their response properties using a mapping and tuning experiment at the start of each recording session. The mapping experiment determined RF extent and the region at which visual stimulation elicited the highest response. This was done by manually moving a static dot pattern on the monitor using mouse control. If the RF size matched the size of an average MT cell (diameter approx. 5 deg, eccentricity dependent), we continued with the tuning task in which one RDP was placed in the RF. While the monkeys performed a luminance detection task, where a slight luminance change of the centrally presented fixation point had to be detected, the dots of the RDP performed brief coherent linear movements in a pair of opposing directions (e.g., 0/180) at various speeds (2, 4, 8, and 12 deg/s). The direction pair differed on each trial in steps of 30 degrees, presented in random order. Based on the tuning profile of the neuron, which was computed online, units were identified as MT neurons and the preferred direction and speed were defined for the subsequent attentional task. When recording several units simultaneously, stimulus properties were chosen to optimally activate all neurons.
Main experiment with pharmacological manipulations
With the information gained in the mapping and tuning tasks, we generated the main experimental task with stimulus properties targeting the isolated neuron(s).
The main experiment is subdivided into three, possibly repeating, blocks: control, injection, and recovery (see Fig. 1b).
During all blocks, the monkeys performed all attentional task conditions in random order. During the control block, neuronal activity was measured in the absence of pharmacological influence. After sufficiently many hit trials (at least 11 repetitions of each condition), a specific amount of a substance was injected every minute, or twice a minute, using pressure injection (gray-shaded area in Fig. 1b). In total, 2 different substances were used in this study: acetylcholine (general agonist) and scopolamine (muscarinic cholinergic antagonist). As a control, saline was injected. Only one substance was used per recording session.
A recovery block followed the injection block, in which no substance was injected. In this block, the neurons’ activity should recover, if affected during the injection block, to the same value as in the control period. In most of the experiments, one full cycle, covering control, injection, and recovery block, was performed. In other rare cases, an additional or even a third cycle of pharmacological manipulation was recorded.
Data analysis
With our custom-made software, we were able to analyze the behavioral data online during the recording session, as well as the spike train of the isolated cells. This helped us gain an initial impression of the data quality while recording. Final data analysis was performed offline using custom scripts written in MATLAB (R2016b), after offline sorting of the recorded neural data.
For the main analysis, only the response to the preferred direction of each neuron was used. Firing rates were averaged in an analysis window of 300–800 ms after RDP onset for every trial.
We compared firing rates in the absence (initial control block) and presence of a drug (injection block/s). To determine statistical significance, two-sided non-parametric tests (Wilcoxon signed-rank for paired data or a test of one sample against a median of 0; rank-sum for comparing two independent samples) were used unless otherwise indicated.
The trials recorded before the first injection provided control data. The injection block started with the first trial occurring > 150 s after the first injection and ended 150 s after the last injection (see Fig. 1b).
Inclusion criteria
The first inclusion criterion examined data quality for each recorded neuron, independent of the effects of injection and attention. Here, we only included data that contained enough trial repetitions during the injection block(s) for every task condition (a minimum of 3 trials). Additionally, recorded cells had to respond with a certain strength to the preferred stimulus (minimum firing rate of 7 spikes/s). A responsiveness check was also performed comparing the sensory conditions for preferred and null direction and fix-only during the control block. Cells were included in further analysis if they showed significant differences between the three fixation conditions (Kruskal-Wallis test, p < 0.05).
A second criterion was applied to define the subsets of cells showing a significant effect of each injected substance. We compared the distribution of firing rates in the preferred direction sensory condition (attend-fix) in the control block with those in the injection block (Wilcoxon rank-sum test, p < 0.05).
Only one substance was injected in each recording session. One hundred thirty well-isolated MT single units of two monkeys, recorded from both hemispheres, fulfilled the first inclusion criteria when using the muscarinic antagonist scopolamine. Twenty-eight units were used for saline control. Ninety units were used to analyze the effect of general agonist acetylcholine.
Quantifying modulation
We measured the effects of attention and injection in a fixed time window during the sustained response of the neuron to the moving stimulus in its receptive field (300–800 ms after the onset of the RDPs). An injection modulation index (IMI) was used to quantify the influence of injection on firing rates. It was defined as the difference in average firing rate, divided by the sum.
$$ \mathrm{IMI}=\frac{R_2-{R}_1}{R_2+{R}_1} $$
where R1 is the firing rate in the control block and R2 in the injection block. Therefore, a positive IMI indicated a response enhancement due to injection. The IMI was calculated separately for all conditions: attend-fix, attend-in, and attend-out.
In order to test for injection effects on attentional modulation, an attentional modulation index (AMI) was calculated and compared for control and injection blocks.
The AMI was used to measure the effect of attention on firing rate and was defined as:
$$ \mathrm{AMI}=\frac{Q_2-{Q}_1}{Q_2+{Q}_1} $$
where Q1 is the respective firing rate in attend-out trials and Q2 in attend-in trials. Positive values indicate an increase in response by attention, whereas negative values denote suppression.
The AMI and the IMI were also used to define the percentage modulation in firing rate due to attention or injection.
$$ \mathrm{percAMI}=\frac{2\times \mathrm{AMI}}{1-\mathrm{AMI}}\times 100 $$
The percentage change for the IMI was calculated in the same manner as for the AMI.
Test for potential injection effects
In order to exclude an influence of the injection process per se on the neuronal firing rate, we injected saline as a control substance, following the same protocol as the other injections. As depicted in Additional file 1: Figure S1, 28 of those neurons were included in the analysis (see criteria above). This group of neurons was not significantly affected by saline injection during the fixation condition (fixation condition vs. 0; N = 28, p = 0.095, W = 129).
Additionally, we could exclude the influence of the injection process on attentional enhancement, as attention modulation was not significantly changed with saline injection (N = 28, p = 0.479, W = 171, inset histogram in Additional file 1: Figure S1b).