Cholinergic optimization of cue-evoked parietal activity during challenged attentional performance

Abstract

The detection of salient or instrumental stimuli and the selection of cue-evoked responses are mediated by a fronto-parietal network that is modulated by cholinergic inputs originating from the basal forebrain. Visual cues that guide behavior are more strongly represented in the posterior parietal cortex (PPC) than are similar cues that are missed or task-irrelevant. Although the crucial role of cholinergic inputs in cue detection has been demonstrated by lesion studies, the role of PPC neurons in the cholinergic modulation of cue detection is unclear. We recorded extracellular spikes from PPC neurons of rats performing a sustained attention task, before and after selective removal of cholinergic inputs to the recording site. Visual cues that were subsequently detected evoked significant increases in the PPC firing rate. In the absence of cholinergic input, the activation of PPC neurons by detected cues was greatly diminished. When a visual distractor was introduced during task performance, a population of PPC neurons selectively responded to the distractor. As a result of cholinergic deafferentation, distractor-related neuronal activity was enhanced, and the detection-related activity was further suppressed. Thus, in deafferented subjects, the distractor lowered the signal-to-noise ratio of cue-evoked responses. This impairment in cue-evoked neuronal activity may have mediated the increased response latencies observed for detected cues in the presence of the distractor. Additional experiments demonstrated that the effects of cholinergic deafferentation were not confounded by extended practice or electrode depth. Collectively, these findings indicate that cholinergic inputs to PPC neurons amplify cue detection, and may also act to suppress irrelevant distractors.

Fig. 1

Depiction of task performance, timeline of main experimental events, and behavioral results. (A) At the start of each trial, cues (illuminations of a panel light at 25, 50 or 500 ms) are either presented (top sequence) or not (non-signal trial; bottom sequence). A 250-ms tone is initiated 1 s later, which opens a 4-s operant window. Animals were required to press one lever to report a hit and the other lever to report a correct rejection to receive a reward. A response or an omission (no response in 4 s) initiates a variable intertrial interval (ITI) (10 ± 3 s) for the next trial. (B) Single unit activity was recorded from task-performing animals for two standard sessions (S) and a distractor session (D; house light flashes at 0.5 Hz for the middle 12-min block of trials). After recordings had been collected from 15 sessions with intact PPC neurons, a 0.5-L bolus of the selective cholinotoxin 192-IgG saporin (SAP) (n = 5) or Dulbecco's saline (n = 2) was infused near the recording site of animals. Recordings were then collected from 15 post-infusion sessions (10 standard, five distractor) for each animal. (C and D) Behavioral performance as a function of signal duration, distractor and trial block. (C) Plots showing the relative number of hits during standard and distractor sessions. Cue detection was signal duration-dependent (F_2,44 = 254.52, P < 0.05), and was hampered by the distractor (F_2,44 = 6.5, P < 0.05). (D) The bar graphs show that the distractor (Dist) reduced the number of correct rejections relative to standard session performance (Stand) in lesioned and saline-infused animals (Dist; F_2,44 = 18.82, *P < 0.05). Solid bars represent performance prior to SAP or saline infusions, and shaded bars represent post-infusion data. Cholinergic denervation did not affect accuracy on either signal or non-signal trials. CR, correct rejection; FA, false alarm.

Fig. 2

Cholinergic deafferentation delays hit latency during distractor sessions. During standard sessions, lesioned subjects show an initial, non-significant increase in latency (F_2,44 = 0.26; P > 0.05). During distractor trials, cortically deafferented subjects display a significant increase in response latency during standard performance blocks 1 and 3 relative to saline-infused subjects (F_1,22 = 5.88; P < 0.05). Distractors also delayed hit latency on both groups (F_2,44 = 6.70; P < 0.05). SAP, 192-IgG saporin.

Fig. 3

Electrode path and 192 IgG-saporin-induced acetylcholinesterase (AChE)-positive fiber loss within the posterior parietal cortex (PPC). (A) Schematic of a coronal section through the level of the PPC (4.3–4.5 mm posterior to bregma), illustrating the final recording sites of the recording electrodes in the left or right hemispheres. Rats were infused with 192-IgG saporin (SAP) (n = 5, red circles) or saline (n = 2, green squares). (B) Photomicrograph (×4) of the electrolytic lesion at the final recording site of Nissl-stained PPC. Electrolytic lesions from each animal were within PPC layers III–V of the left hemisphere. (C) Restricted loss of AChE-positive fibers to the PPC near the final recording site in the left parietal cortex. SAP lesions reduced the density of AChE-positive fibers by 75% as compared with the contralateral cortex (F_1,35 = 119.605, P < 0.05). (D) Higher-magnification (×20) photomicrograph of AChE-positive fibers from the same site as in C. (E) High-magnification photomicrograph of the contralateral AChE fibers from the PPC of the same animal as in C.

Fig. 4

Behaviorally relevant cues elicit increases in firing rate from posterior parietal cortex (PPC) neurons. (A) Raster plot and histogram of a single PPC neuron during the four response types. Clockwise from top left: correct detection of a signal (Hit), incorrect rejection of a signal (Miss), correct rejection (CR) of non-signal trials, and incorrect detection of non-signal trials [false alarms (FAs]. The colored circles represent the onset of the visual signal; the black line represents the onset of the tone, opening the operant window on all trials. Symbols following the black line represent the behavioral response on each trial (green triangles, Hit; inverted green triangle, CR; orange diamond, Miss; orange square, FA). Rasters are organized by trial number, with the first trial on the top. The number of trials in each session is indicated in parentheses at the top of each graph. (B) Neural responses from 92 PPC neurons prior to infusions. The response to the visual cue initially peaks at 220 ms following cue presentation, and activation remains sustained until the hit response. The cue-evoked response was greater on hit trials than on miss trials. Smaller curves on the bottom represent the reaction time distributions for correct responses on cued and non-signal trials. False alarm and miss trials have a similar distribution and latency as CRs, and have been omitted for clarity. (C) Neural responses from 25 cue-driven PPC neurons following saline infusions exhibit a similar, detection-specific pattern.

Fig. 5

Cue-driven activity at 25-ms, 50-ms and 500-ms signal durations, normalized to the pre-signal average firing rate. The top panels represent the cue-evoked firing rate of all hit-related neurons for each standard session condition; the bottom panels represent the firing rate of hit-related neurons during distractor conditions. (A) Prior to saline or saporin infusions 500-ms cues elicited a larger neurophysiological response (see text). (B) Cue-driven activity in standard sessions following cholinotoxin lesions. All durations evoked a similar response from cue-driven neurons (F_2,122 = 2.08, P > 0.05). (C) Cue-driven activity in standard sessions following saline infusions (F_2,48 = 1.14, P > 0.05). (D) Pre-lesion cue-driven activity in distractor sessions (F_2,98 = 0.05, P > 0.05). (E) Cue-driven activity in distractor sessions following cholinotoxin lesions (F_2,34 = 0.34, P > 0.05). (F) Cue-driven activity in distractor sessions following infusions of saline (F_2,24 = 1.35, P > 0.05). L, signal light; T, tone.

Fig. 6

Cholinergic lesions increase the proportion of posterior parietal cortex (PPC) neurons responsive to the visual distractor. (A) Population of neurons that are significantly activated when the distractor light is off. (B) Distribution of significant correlates for pre-lesion and post-lesion sessions. The distractor activates 40/208 neurons prior to infusion of 192-IgG saporin. The peak activation of this population took place 20 ms after the distractor light was turned off. Following cholinergic deafferentation, a greater proportion of PPC neurons are activated by the off phase of the distractor (52/167, 31%; (χ² = 20.32, P < 0.05), and fewer neurons are activated by the signal light. In both graphs, neurons that have mixed correlates are activated by both the signal and the distractor.

Fig. 7

Lesions caused by 192-IgG saporin (SAP) showed a trend towards reducing the signal-to-noise ratio (SNR) of cue-evoked posterior parietal cortex (PPC) neurons in distractor blocks. (A) Prior to the administration of SAP, cue-driven PPC neuron activity is equivalent in both the distractor and standard trial blocks. The SNR [SNR = R_stim/(R_stim+R_spont)] was calculated from each of the 36 cue-driven neurons during each trial block; the SNR of each neuron in standard blocks is plotted with respect to the SNR during the distractor block. The cue-driven activity was equivalent in both trial blocks (Z = −0.833; P > 0.05, Wilcoxon). (B) Following infusions of SAP, the SNR of 18 cue-driven neurons in standard trial blocks remained elevated, but the SNR was relatively lower in the distractor block, with a trend towards significance (Z = −1.851, P = 0.06). (C) The SNR was equivalent for both the standard and distractor trial blocks following saline infusions (Z = −1.306; P > 0.05).

Fig. 8

Detection-related PPC neurons respond similarly on hit and miss trials following cholinergic lesions. (A, left) Stimulus-locked population peri-event time histograms (PETHs) for hit and miss trials (20-ms bins, Gaussian filtered over three bins) of 36 cue-evoked neurons from pre-lesion recordings. Peak activation of these neurons on hit trials occur at an average of 220 ms following the cue, and the firing rate remains elevated through the 1-s delay and response. (A, right) The average firing rate of each neuron during the 1-s epoch following the cue, with hit trials plotted against miss trials; 28/36 neurons (green dots) have a significantly higher firing rate on hit trials (all P < 0.05). Only 2/36 neurons (red dots) have a significantly higher firing rate on miss trials. (B, left) A stimulus-locked PETH from 18 cholinergically lesioned cue-driven neurons. PPC neurons are activated on both hit and miss trials. (B, right) The average firing rate of each neuron during the 1-s epoch following the cue; 6/18 neurons have a higher firing rate on hit trials than on miss trials, and 3/18 neurons have a higher firing rate on miss trials. (C, left) Following saline infusions, PPC neurons are activated on hit trials, but not miss trials. Right: 10/15 neurons have a higher firing rate in the 1-s epoch following the cue on hit trials than on miss trials.