Evoked Frontal and Parietal Field Potential Signatures of Target Detection and Response Inhibition in Rats Performing an Equiprobable Auditory Go/No-Go Task

The P2 amplitude reflects target detection, not just response activation

We observed an amplification of the P2 amplitude on hit trials as compared to CRs, consistent with the involvement of P2 activity in target detection or initiation of a licking response, which both occur during hit but not CR trials. That we also observed an elevated P2 amplitude on hits compared to FA trials, supports that this amplification cannot solely be accounted for in terms of licking-related activity. Thus, we conclude that neural activity underlying the P2 ERP peak is involved in detection of the target tone. We also observed a non-significant trend toward larger P2 amplitude on hit trials than misses, which is consistent with the target detection interpretation of the P2 function. We did not observe any significant difference between the frontal and parietal P2 amplitudes.

In principle the P2 might have been larger in amplitude on hit trials than CRs or FAs due to the target pitch (3000 Hz) intrinsically (i.e., independent of target detection or attentional state) eliciting a larger P2 response than the distractor pitch (1500 Hz), despite their identical physical intensity. However, as there was no significant difference between the P2 on miss trials (when the target pitch was presented) and either CRs or FAs (when the distractor was presented), significant pitch “tuning” in the P2 response is unlikely. Furthermore, recordings in two control animals during the equally probable passive presentation of the two pitches revealed a significant trend toward a larger P2 amplitude in response to the low pitch as compared to the higher pitch. While these data are not sufficient to establish frequency tuning in the rat P2 component, tuning following the trend in our data would reduce the relative P2 amplification on hit trials that we observed. Consistent with these considerations, an investigation of frequency tuning of the auditory ERP in rats did not report significant tuning of this auditory ERP component (Knight et al., 1985). We thus conclude the P2 amplification we observed on hit trials most likely reflects target detection related neural activity.

This conclusion is comparable to findings from surface recordings in rats (Shinba, 1997; Ahnaou et al., 2018) which show a P2 with a parietal maximum that was larger in response to target than distractor tones in a head-fixed auditory active oddball task. This refers to a go/no-go task in which the targets are relatively rare. Similar observations have been reported by others in anterior cingulate cortex and hippocampus, but without statistically comparing P2 amplitude on different trial types (Hattori et al., 2010). However, in another study, skull recordings at the vertex failed to show a significant difference between P2 amplitudes in response to targets and standards in an auditory active oddball task (Sambeth et al., 2003).

Regardless, as target trials were rare in these previous studies, oddball amplification of responses to rare tones was conflated with effects of target detection processes during the active task. Indeed, frontal and parietal cortical rare-tone amplification effects have been reported in rats at multiple ERP peaks including the P2 (Imada et al., 2013). In contrast to these previous studies, the target and distractor stimuli were presented with equal probability in our auditory go/no-go task, so a rare-tone amplification effect is ruled out. Thus, to our knowledge the present results are the first to specifically implicate the rat P2 in target detection as distinct from response activation processes during an auditory go/no-go task.

The P2 enhancement we observed during hit trials in rat frontal and parietal cortex appears analogous to the go-P2 around 200 ms post-stimulus reported in recent equiprobable auditory go/no-go ERP experiments in humans (Borchard et al., 2015) and other auditory go/no-go studies (Gajewski and Falkenstein, 2013). The human frontal-parietal P2 peak was larger during go trials and could thus be associated with response activation or target detection as we found in the rat.

The N2 negativity does not index response inhibition but may reflect error processing

In our dataset the N2 amplitude was not consistently associated with active response inhibition on no-go trials. We instead found that the N2 amplitude is larger on miss than CR trials, which might indicate error-related cortical processing. This interpretation is dubious in our dataset because our rats were relatively proficient at the task, resulting in fewer incorrect than correct trials in each session, leading to a significantly greater variance in our incorrect trial ERPs than correct trial ERPs. Greater variability could be reflected as a bias toward systematically larger peaks in the miss trial ERPs. Thus, while our results are consistent with the possibility that the N2 negativity reflects processing or monitoring of behavioral errors, we cannot rule out a contribution from bias due to different-sized samples of each trial type.

The N2 amplitude has been associated with error-processing in some human ERP experiments. In this context the component is known as the error-related N2 or error negativity (Ne; Falkenstein et al., 1991; Scheffers et al., 1996; Botvinick et al., 1999; Kok et al., 2004). The N2 response we observed in rats may be analogous to these error-related negativities described in human ERPs. Future experiments may be able to address this issue more decisively if more difficult tasks are used to elicit greater numbers of error trials.

In humans, multiple go/no-go ERP studies using visual or auditory stimuli have reported a frontal-central no-go N2 around 200–400 ms post-stimulus that is enhanced during no-go relative to go trials (Pfefferbaum et al., 1985; Jodo and Kayama, 1992; Sasaki et al., 1993; Watanabe et al., 2002; Bekker et al., 2005a,b; Kaiser et al., 2006; Smith et al., 2007; Borchard et al., 2015). Because the no-go N2 was present even when targets on go trials were merely counted mentally rather than their presence being reported with an overt behavioral response (Pfefferbaum et al., 1985; Smith et al., 2008), it may reflect “cognitive control” (Folstein and Van Petten, 2008) or “cognitive inhibition” in auditory (Smith et al., 2008) and visual go/no-go tasks (Gajewski and Falkenstein, 2013), rather than motor inhibition. Again, our results do not support that such a process occurs in the rat brain, at least in the context of equiprobable target and distractor stimuli.

The (earlier) P3E peak does not reflect target detection, consistent with a previously identified role in deviance detection

Previous rat work reporting a P3 response around 200%, 250 ms (Yamaguchi et al., 1993; Imada et al., 2013), suggest this component reflects novelty or deviance analogously to the human P3a (Linden, 2005; Polich, 2007). However, other rat studies have reported a go-related P3 with peak latencies overlapping the novelty-related P3 but ranging to later times up to or exceeding 400 ms post-stimulus (Shinba, 1997; Sambeth et al., 2003; Hattori et al., 2010). We therefore defined separate early and late P3 peaks.

Neither stimulus in our experiments was rare compared to the other, so we expect no novelty, deviance, or mismatch-related contribution to the P3E in our experiments. This task design allowed us to attribute a role in target detection to the P2. In contrast to the P2, we did not find support for the P3E having a role in target detection. Rather, we noted that the P3E amplitude on miss trials was significantly larger than on CR trials. While this might reflect error processing after failed go trials, this interpretation is subject to the same concern as the previously discussed N2 error-related results, as the miss ERPs are based on fewer trials than the CR ERP from the same session.

The (later) P3L peak does not index target detection, response inhibition or production

We observed a significantly larger P3L on misses than FAs, arguing against a role of the P3L in response production. The P3L was also significantly larger on misses than CRs, and not significantly different on hits and misses, arguing against a role of the P3L in target detection. As previously mentioned, interpreting the larger P3L amplitude on miss trials as an indication of error-processing is subject to the concern that error trials were fewer and their associated ERPs more variable, likely resulting in a bias toward larger peak amplitudes on incorrect trials.

Previous auditory ERP studies in the rat have generally presented ERPs showing a larger target P3 than distractor P3, which in some cases may not be present at all (Shinba, 1997; Sambeth et al., 2003; Hattori et al., 2010; Ahnaou et al., 2018). These studies used oddball paradigms meaning that target tones were rare compared to distractors. Their results are consistent with the P3 playing a role in target detection or response production, but since the target tones were rare compared to distractors in these studies, interpretation of the P3 is confounded by potential oddball responses to the rare target tones.

Similarly, multiple human studies (Pfefferbaum et al., 1985; Rockstroh et al., 1996; Ochoa and Polich, 2000) describe a P3 peak around 300 ms post-stimulus with a parietal or parietal-central maximum that is larger during target than standard trials in oddball tasks. In equiprobable go/no-go tasks this P3 has been associated with response production (Pfefferbaum et al., 1985; Borchard et al., 2015), as well as target detection in auditory mental counting experiments without direct motor responses (Picton and Hillyard, 1974; Wronka et al., 2008).

Our results do not support a role for either the early or late P3 peak in target detection. Rather they suggest that the larger P3 on target trials noted in previous rat work could be due to an oddball amplification of the target response as opposed to target detection. Our P3 results are consistent with a study of parietal ERPs in rats performing a visual sustained attention task, which found a “P300” response (peaking before ∼250 ms post-target onset) that was sensitive to target stimulus duration but not to detection (i.e., no hit-miss difference) or attentional load (Broussard and Givens, 2010). They concluded that the parietal P3 reflected “sensory aspects of the target, and not detection per se.”

General comparison to human ERPs

In terms of comparing the phenomenology of rat auditory ERPs to humans’, Sambeth et al. (2003) considered the rat P1, N1, P2, N2, and P3 peaks as corresponding to human ERP components simply according to their order of occurrence and polarity, and found the peak latencies in the rat to be 1.8 times earlier in the rat as compared to humans in roughly matched rat and human auditory active oddball paradigms—except for the P3 which occurred at roughly the same latency in rats and humans (350–380 ms in that study). On this basis they suggested the earlier peaks reflect “sensory processes” while the P3 reflects a more “cognitive…elaborative processing of stimuli” that is amplified by the expectation of reward in the active task. This is consistent with the demonstration of rare-tone response enhancements of the P3E in rats that could not be accounted for by simple stimulus-specific adaptation mechanisms (Imada et al., 2013), suggesting some more sophisticated memory comparison at the P3E between rare oddball tones and the recent history of standard tones.

As discussed previously, the human ERP literature suggests the involvement of the P2 and P3 ERP components in target detection and response production, and potential involvement of the N2 and P3 components in response inhibition. While our recordings from rat brain support the role of the P2 peak specifically in target detection, our results do not support a major role of the P3 in target detection. Rather, our results are more consistent with those of Broussard and Givens (2010) who reported a parietal P3 response in a visual task that was not modulated by attentional load and was virtually the same on hit and miss trials. Thus, despite the identification of P3 responses in rat brain that share some properties with their human analogues, the rat P3 responses do not appear to share the role in target detection or response production that has been reported for the human P3.

Similarly, our results do not reveal any obvious analog in the rat of no-go N2 or no-go P3 responses described in some human studies. We cannot decisively conclude at this stage whether these differences represent real species differences or differences in task context. These no-go peaks might emerge in rats, as in humans (Nieuwenhuis et al., 2004), when targets are relatively frequent and distractors are rare, leading to greater priming of the go response and more difficulty or conflict in attempting to inhibit the go response.

Induced oscillatory contributions to ERPs

The similar latency of P3 responses in humans and rats in contrast to the earlier evoked potential components recalls the suggestion that oscillatory brain mechanisms may be tuned to maintain a similar set of frequencies across species from rodents to humans (Buzsáki et al., 2013), and raises interesting questions about how stimulus-induced oscillations may be related to ERP components (Broussard and Givens, 2010; Ahnaou et al., 2018). Our spectral analysis found that post-stimulus induced low-frequency power (1–7 Hz) was greater on hit trials than CRs, especially in frontal cortex, whereas power between 8 and 14 Hz was suppressed on hit relative to CRs. We also calculated average coherence between frontal and parietal electrodes as a function of time during each trial, and observed broadband coherence induced on hit trials as compared to CRs or misses. The induced low-frequency power and broadband interarea coherence on hit trials could reflect detection-related activity but are likely to also reflect rhythmic licking for water.

Comparing miss trials to CRs and hits with our spectral analysis revealed a band near 7 Hz at which power was particularly elevated on miss trials, even before the onset of the tone. Since we observed greater N2, P3E, and P3L amplitudes on miss trials compared to CRs, it is possible that our P3 and N2 ERP results reflect error-related oscillatory activity in the θ range as has been reported in the context of a delayed reaction-time task (Narayanan et al., 2013; Laubach et al., 2015), particularly since our P3L definition allowed later peaks (up to 1000-ms latency) than most previous studies (typically not later than 500-ms latency). However, our observation of elevated θ-power before the unpredictably timed tone on miss trials suggests instead a spontaneous state of reduced attention that predisposes the animal toward failing to respond to, i.e. missing, the target tone. This is consistent with results from a study of cortical oscillatory dynamics in rats performing a simple auditory detection task with unpredictably timed target tones (Herzog et al., 2014). However, that latter study found that both increased low-frequency power and interarea (frontal-parietal) coherence predicted miss trials, whereas our pre-stimulus result was specific to the local synchronization (i.e., power) and did not appear in the interarea coheregrams. This difference between studies may reflect a heightened state of sustained attention during our more complicated discrimination task, resulting in less global low-frequency synchronization (i.e., interarea coherence) even during the animal’s least attentive moments during a session, just before miss trials.