Selective Increase of Auditory Cortico-Striatal Coherence during Auditory-Cued Go/NoGo Discrimination Learning

Abstract

Goal directed behavior and associated learning processes are tightly linked to neuronal activity in the ventral striatum. Mechanisms that integrate task relevant sensory information into striatal processing during decision making and learning are implicitly assumed in current reinforcement models, yet they are still weakly understood. To identify the functional activation of cortico-striatal subpopulations of connections during auditory discrimination learning, we trained Mongolian gerbils in a two-way active avoidance task in a shuttlebox to discriminate between falling and rising frequency modulated tones with identical spectral properties. We assessed functional coupling by analyzing the field-field coherence between the auditory cortex and the ventral striatum of animals performing the task. During the course of training, we observed a selective increase of functional coupling during Go-stimulus presentations. These results suggest that the auditory cortex functionally interacts with the ventral striatum during auditory learning and that the strengthening of these functional connections is selectively goal-directed.

Keywords: Mongolian gerbil; auditory cortex; avoidance learning; discrimination learning; field-field coherence; functional coupling; shuttlebox; ventral striatum.

Figure 1

Representation of coherency as phasor in the complex plane. Instead of the complex-valued coherency S, often only its magnitude is considered (represented by the length of S, then called “coherence” and here denoted “mCoh”). Note that volume conduction alone, adding a real vector component V (red), can already cause an increase in mCoh. In this case (case I), the mCoh increase would be concomitant with a decrease of the phase angle θ of S. By separate analysis of the real and imaginary part of the coherency it is possible to identify mCoh increases based on concomitant increases of the real and imaginary part (case II). In the present study only coherence increases matching case II were accepted as indicating an increase in neuronal coherence free of volume conduction.

Figure 2

Gerbils learned to discriminate FM-direction within 5 training sessions. Shown are the individual learning curves with hit (red curves) and false alarm-responses (black curves). The last plot (gray background) displays the average over all animals with SEM. Animals could attain 30 hits/false alarms per session.

Figure 3

Cortico-striatal projections originated in layer V/VI of the primary auditory cortex and within the anterior and posterior auditory fields. Exemplary histological slices of retrogradely transported nanobeads (left) and NeuN staining (right). Scale bars left column 250 μm, right column 500 μm. AI, primary auditory cortex; DP, dorso-posterior field; AAF, anterior auditory field; CPu, caudate putamen; HCF, hippocampal formation; roman numbers I–IV, cortical layers; wm, white matter; r, rostral; m, medial.

Figure 4

Baseline cortico-striatal coupling remained stable over the course of training. Baseline magnitude and imaginary coherence differed in their frequency-dependence. (A, B) Exemplary spectra of both coherency measures during pre-session period in one animal. (C) Baseline magnitude coherence was independent from frequency range. Shown are grand averages over sessions and animals. Error bars are SEMs. Gray shaded area marks values within the 95 percentile of the bootstrapped mCoh values. (D) Pre-session imaginary coherence was significantly increased in the low frequency range (4–10 Hz). Gray shaded area represents values within the two tailed 97.5 percentile boundaries of a bootstrapped iCoh distribution. (E) Grand average coherency was constant over training sessions for all frequencies analyzed.

Figure 5

Stimulus presentations significantly modulated magnitude and imaginary coherence during trials. (A,C) Average magnitude coherence spectra of one session from the same animal as in Figure 4; trial conditions (CS+ and CS−) were pooled. Black bars represent stimulus onsets and duration. (B,D) Same display as in (A,C) for the iCoh. (E,F) Modulation indexes were significantly increased in the low frequency range 4–10 Hz for mCoh (E) and 4–8 Hz iCoh (F). Depicted are grand averages with SEMs. Gray shaded areas mark 99 percentile boundary of a bootstrapped distribution of modulation indexes from pre-session periods for both coherency measures.

Figure 6

Auditory discrimination learning specifically increased functional cortico-striatal coupling. (A) mCoh increases from early to late training sessions for CS+ presentations were detected in all frequency ranges from 4 to 40 Hz. Shown are the average differences from early to late sessions of all animals. ^*p < 0.05, ^**p < 0.01, post-hoc paired t-tests. Shaded rectangle shows values displayed in (B). (B) Onset mCoh during CS+ trials increased from early to late training sessions (red lines), while presentations of CS– tones did not alter onset mCoh values during training (black lines). Shown are the average z-transformed mCoh values of the 8 Hz frequency band in early and late sessions of individual animals (n = 6). (C) iCoh values significantly increased in the 4–10 Hz frequency range for CS+ presentations. CS− iCoh values were not further analyzed, as mCoh values were not changed during training.

Figure 7

Different motor responses did not cause the increase in CS+ onset coherencies. (A) Onset magnitude coherence (mCoh) was not differentially changed during CS+ presentations when split into shuttling (hit) and staying (miss) responses. Shown are grand averages for late and early sessions of miss trials (white dots) and hit trials (black dots) in the 8 Hz frequency band. (B) Differences between hit and miss trials from early and late training sessions were not frequency dependent. Values ranged around zero for all frequencies except 4 Hz; here the difference was negative, indicating higher mCoh values for non-jump responses. Error bars represent SEM.

Figure 8

Cortico-striatal spike-field coherence was elevated during auditory stimulation. Auditory stimulation led to significantly increased spike-field coherence in nearly all frequency ranges from 1 to 100 Hz (blue line). Gray area represents 99 percentile-bound distribution of bootstrapped pseudo spike- field coherences from pre-session recordings. Average pre-session spike-field coherence is represented by the magenta curve. Inlet shows recorded spike shapes; scale: x: 0.5 ms, y: 0.15 mV.