Attentional Processing in the Rat Dorsal Posterior Parietal Cortex

Abstract

The human posterior parietal cortex (PPC) is known to support sustained attention. Specifically, top-down attention is generally processed in dorsal regions while bottom-up regulation occurs more ventrally. In rodent models, however, it is still unclear whether the PPC is required for sustained attention, or whether there is a similar functional dissociation between anatomical regions. Consequently, the aim of this study was to investigate the contribution of the rodent dorsal PPC (dPPC) in sustained attention. We used the five-choice serial reaction time task (5CSRTT) and compared rats with neurotoxic dPPC lesions to sham operated rats. We found that rats with dPPC lesions were less accurate and took longer to make correct choices, indicating impaired attention and reduced processing speed. This effect, however, was limited to the first few days of post-operative testing. After an apparent recovery, omissions became elevated in the lesion group, which, in the absence of reduced motivation and mobility, can also be interpreted as impaired attention. In subsequent challenge probes, the lesion group displayed globally elevated latency to make a correct response, indicating reduced processing speed. No differences in premature responses or perseverative responses were observed at any time, demonstrating that dPPC lesions did not affect impulsivity and compulsivity. This pattern of behavior suggests that while intact dPPC supports goal-driven (top-down) modulation of attention, it likely does not play a central role in processing stimulus-driven (bottom-up) attention. Furthermore, compensatory mechanisms can support sustained attention in the absence of a fully functioning dPPC, although this occurs at the expense of processing speed. Our results inform the literature by confirming that rodent PPC is involved in regulating sustained attention and providing preliminary evidence for a functional dissociation between top-down and bottom-up attentional processing.

Graphical Abstract

Safe and effective interaction with our environment requires maintaining focus on specific goals while also monitoring our surroundings for behaviorally relevant changes. This skill involves balancing top-down (goal directed) and bottom-up (stimulus driven) attentional processes. In top-down attention, sensory information is selected based on internal goals, while in bottom-up attention, we orient to salient stimuli such as a very loud noise or bright color (Connor, Egeth, and Yantis, 2004). An emerging view is that the posterior parietal cortex (PPC) is vital for the integration of top-down and bottom-up attention and that each of these functions relies on different sub regions in primates (Ciaramelli, Grady, and Moscovitch, 2008; Corbetta, Patel, and Shulman, 2008; Corbetta and Shulman, 2002; Shomstein, 2012). What is yet to be determined is whether this functional dissociation is also observed in the rat PPC. Until recently, this gap in knowledge was compounded by the lack of consensus on stereotactic coordinates for homologous PPC areas. Considering the vital role of rodent models, the current study elucidates the role of the dorsal portions of PPC in both bottom-up and top-down attention.

The PPC is considered a higher order association area where multiple streams of sensory, motor, proprioceptive, and vestibular input converge (Whitlock, 2017). One of the more established roles of the PPC is in spatial awareness, which is supported by multiple lines of evidence. Classic lesion literature, for instance, shows that damage to this area can cause profound sensory neglect (Bates and Ettlinger, 1960; Corbetta and Shulman, 2011; Lamotte and Acuña, 1978; Vallar and Calzolari, 2018). In addition, neural activity in the PPC is associated with behaviors that require focused visuospatial attention (Arrington, Carr, Mayer, and Rao, 2000; Duhamel, Colby, and Goldberg, 1992; Leinonen, Hyvärinen, Nyman, and Linnankoski, 1979; Merriam, Genovese, and Colby, 2003; Taira, Mine, Georgopoulos, Murata, and Sakata, 1990; Yang, Jacobson, and Burwell, 2017). Finally, the role of attentional processes in memory retrieval has also been established (Ciaramelli et al., 2008). Taken together, these lines of evidence indicate that the PPC is an association area that incorporates spatial and sensory information with the purpose of directing attention.

As in primates, the rodent PPC appears to be crucial for mediating sensory input, especially in the visual modality (Akrami, Kopec, Diamond, and Brody, 2018; Broussard and Givens, 2010; Burcham, Corwin, Stoll, and Reep, 1997; Kolb and Walkey, 1987). It is also central in attentional processing, with involvement in orienting behavior (Tees, 1999), contralateral inattention (Burcham et al., 1997), memory-reliant foraging (Espina-Marchant, Pinto-Hamuy, Bustamante, Morales, Robles, and Herrera-Marschitz, 2006), visuospatial sustained attention (Broussard, Sarter, and Givens, 2006; Husain and Nachev, 2007; Yang, Dokovna, and Burwell, 2021; Yang et al., 2017) and attentional shifting (Fox, Barense, and Baxter, 2003). Anatomically, the rat PPC is located between somatosensory and visual cortex and receives similar input from both cortical and thalamic areas (Bucci, 2009). Taken together, these data suggest that rodent PPC may be homologous to primate PPC, performing a similar task of processing sensory information to support attention. What is not yet clear, is whether PPC subdivisions function similarly in primates and rodents.

One remarkable aspect of visuospatial attention in the primate PPC is that it is anatomically and functionally dissociated. In general, top-down processing occurs dorsally in the superior parietal lobule (SPL; Brodmann areas 5 and 7), while bottom-up attention is processed more ventrally in the inferior parietal lobule (IPL; Brodmann areas 39 and 40) and temporo-parietal junction (Ciaramelli et al., 2008; Corbetta and Shulman, 2002; Shomstein, 2012). The rodent PPC is also sub-divided into regions: dorsal (dPPC) and ventral (vPPC). These areas are anatomically delineated by differing cytoarchitecture and thalamic connectivity, which is homologous to primate PPC connectivity (Olsen, Ohara, Iijima, and Witter, 2017; Olsen and Witter, 2016). In addition, it is possible that dPPC and vPPC serve different functions in visuospatial attention (Olsen and Witter, 2016). For example, in vivo electrophysiology shows a greater proportion of stimulus selectivity in vPPC compared to dPPC cells when rats are performing a visuospatial task, as well as earlier selectivity (Yang et al., 2021). These lines of evidence suggests that rodent PPC may also be functionally dissociated and that vPPC, like the primate IPL, may have a more prominent role in bottom-up visual attentional processing than dPPC.

One specific aspect of visuospatial attention that requires further investigation is sustained attention. Single cell recordings confirm that PPC neurons are active during a sustained attention task (Broussard et al., 2006), especially when ignoring task-irrelevant (i.e. distracting) stimuli (Broussard and Givens, 2010). However, although neural correlates are evident, it is still unclear whether PPC cells are actually driving sustained attention. Evidence from cholinergic inactivation of dPPC inputs shows that sustained attention is not impaired when measured by a 5-choice serial reaction time task (5CSRTT) (Maddux, Kerfoot, Chatterjee, and Holland, 2007), a visuospatial attention task that measures sustained attention (Bari, Dalley, and Robbins, 2008). Cholinergic inactivation does impair surprise-mediated enhancement of attentional learning in a serial conditioning task during encoding, consolidation, and retrieval (Bucci, Holland, and Gallagher, 1998; Schiffino, Zhou, and Holland, 2014). Broad neurotoxic lesions to the PPC did not affect sustained attention in a 5CSRTT either (Muir, Everitt, and Robbins, 1996). This would appear to suggest that while PPC neurons are active during sustained attention tasks, their contribution is not necessary for sustained attention to occur. These studies, however, were performed before the anatomical boundaries of the rodent PPC were redefined (Olsen and Witter, 2016) and, as such, the extent of damage to dPPC and/or vPPC was unclear. Thus, whether the PPC has a causative role in processing sustained attention is still an open question.

The current study aims to test attentional processing following lesions to the dPPC specifically. To this end, rats with neurotoxic lesions to the dPPC (n=10) were compared to sham operated rats (n=9) on the 5CSRTT. Rats were trained pre-operatively and then tested post-operatively on a standard version of the task and on a series of challenges including reduced stimulus duration times, distracting task-irrelevant auditory stimuli, and variable stimuli onset times. These challenges are intended to increase the cognitive load and can therefore highlight certain task parameters (Amitai and Markou, 2011). Although not their traditional use, these additional measures could help parse out potential differences between top-down and bottom-up attentional processing. Our results will clarify the role of dPPC in visuospatial tasks. Considering evidence showing that the primate dorsal PPC processes top-down sustained attention and considering the improved coordinates for the rat PPC, we expect to find deficits in the 5CSRT task, both in a standard version and in top-down challenge versions, but not in challenges which might rely on bottom-up attention.

Methods

Subjects

Twenty-four adult male Long Evans rats were used in this study (Charles River Laboratories, Boston, MA, USA). Four rats did not reach pre-operative performance criteria and one died of postoperative complications. Therefore, a total of 19 rats were included in the experiments. On arrival, all rats were pair-housed in ventilated cages in a reverse light-dark cycle (lights on at 7 pm and lights off at 7 am) with ad libitum access to food and water. A week before testing, rats were food-restricted and maintained at 85–90% of free-feeding weight. All testing occurred during the dark phase of the light cycle, at approximately the same time each day (starting at approximately 9 am). Beginning one week after arrival, rats were handled at least three times per week for the duration of the experiments. At the time of surgery, all subjects were 4–5 months old and weighed 400–460 g. All procedures were carried out in accordance with NIH guidelines for the care and use of rats in research and protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Providence College and Brown University.

Apparatus

5CSRTT testing was performed in a dedicated room that housed ten operant test chambers (21.6 × 17.8 × 12.7 cm; MED Associates, St. Albans, VT) (Figure 2). Each chamber consisted of aluminum side panels, evenly spaced stainless-steel rods on the floor, and clear Plexiglas© front door, rear wall, and ceiling. Five stimulus-response ports (2.5 × 2.5 × 4 cm deep) were centered on the curved left-side panel, spaced 2.5 cm apart, and raised 2 cm above the grid floor. Each port contained a 3W lamp that provided behavioral cues and infrared phototransistors that detected subject response. A recessed food dipper was centered on the right-side panel 2 cm above the floor. The food dipper was illuminated when pellets were delivered and infrared phototransistors detected head entries into the dipper. A house light (28 V; 100 mA) was located on the left-side panel (centered; 10 cm from the top) that was illuminated during ongoing trials. A computer-automated sound generator was located on the right-side panel and provided programmable auditory stimuli. Each chamber was enclosed in a 62 × 56 × 56 cm sound- attenuating cabinet fitted with an exhaust fan that provided constant air flow and background noise. The apparatus was controlled by Med-PC (MedAssociates, Inc).

Figure 2 – Experimental Setup.

The 5CSRTT was conducted in a sound-attenuating operant testing chamber (2.4 × 30.5 × 29.0 cm) controlled by the MED-PC software package and custom software was written in the Pascal-based, MED-PC notation. A dimly illuminated food cup (food dipper) was recessed in the center of one end wall with a photo beam mounted inside, permitting automated assessment of food dipper behavior. The opposite wall was curved and incorporated five nose poke holes (stimulus response ports) with LED stimulus lights inside. Rats were required to place their nose into the illuminated nose poke hole within the time limit to receive reward. Daily testing sessions consisted of performing 100 trials or 30 minutes of testing, whichever occurred first.

Surgery

After initial training on the 5CSRTT, rats were assigned to either the sham or lesion group according to their performance on the last five sessions of training. Assignments were counterbalanced across all principle dependent measures (e.g., accuracy, omissions, latencies, perseverative errors) and also for operant chamber placement and litter origin. Each lesioned rat was paired with a sham that had similar percent accuracy and omissions, and surgery was performed on the same day for each rat pair. One rat died of postoperative complications leaving a total of 9 shams and 10 lesions. Rats were anesthetized with isoflurane and secured in a stereotaxic frame. The incisor bar was adjusted such that bregma and lambda were in the same horizontal plane (±0.2 mm). Craniotomies were made using a dental drill and the dura mater removed to allow insertion of a 28-gauge, 1 mL capacity Hamilton^® syringe (Sigma-Aldrich) into the target brain region. In the PPC lesion group, five sites were targeted bilaterally (PPC1–5). Bregma was measured using the syringe at an angle of 0° for dPPC1, dPPC2, and dPPC3, and 20° for vPPC4 and vPPC5. Anterior to posterior (AP) and medial to lateral (ML) coordinates were calculated relative to bregma, and dorsal to ventral (DV) coordinates were calculated relative to the top of the skull. The stereotaxic coordinates (AP, ML, DV) were developed using Olsen and Witter (2016) and (Paxinos and Watson, 2013), and modified based on the spread seen in 4 pilot animals; −3.6 mm, ±2.4 mm, −0.5 mm for PPC1, −3.6 mm, ±3.8 mm, −0.5 mm for PPC2, −4.3 mm, ±4.2 mm, −0.5 mm for PPC3, −5.0 mm, ±4.8 mm, −0.95 mm for PPC4, and −5.6 mm, ±5.2 mm, −1.75 mm for PPC5 (Figure 1). Since the rats in this study were much larger (400–460 g) and older than the pilot animals, these coordinates targeted the dPPC only. Neurotoxic lesions were made using NMDA (0.09M in 0.5N sodium hydroxide; Tocris Bioscience, Minneapolis, MN) delivered by pressure injection at 0.1 μL/min for 1 min at PPC1, PPC2, and PPC3, and for 30 s at PPC4 and PPC5. The syringe was left in place for one minute after each injection and then slowly retracted. For sham surgeries, the craniotomy was opened but the syringe was not lowered. For all surgeries, the skin was sutured and rats were allowed to recover for 1 week prior to handling and for an additional week prior to behavioral testing.

Figure 1 – PPC Coordinates.

Extent of lesioned area in the PPC across available literature, and their respective coordinates used to target the PPC. Light grey bars indicate the range of the lesion extent, with the darker spot indicating the average percentage of the PPC lesioned. Previous literature has shown a range of PPC stereotaxic coordinates corresponding to the variable range that has been debated through the history of research into this cortex area, as well as the inclusion or exclusion of the caudal PPC which can be difficult to access. As the boundaries of the rat PPC have recently been clearly delineated using cytoarchitectonic and thalamic connectional criteria (Olsen and Witter, 2016), the coordinates in the current study have aimed to match this new delineation as much as possible.

Behavioral Procedure

Standard Task

In the baseline task, rats performed 100 trials in one daily session. The beginning of a session was indicated by house lights and the release of a free food pellet. When a head entry to the food dipper was detected, a new trial began. At the beginning of each trial, a 500 ms light stimulus was delivered in one of the five stimulus ports. The rat was expected to respond by poking its nose into a stimulus port and was given a 5 second ‘limited hold’ to do so. A ‘correct’ nose poke (NP) into the previously illuminated port was rewarded with a dispensed food pellet (5TUM grain-based 45 mg dustless pellets, Test Diet) that was indicated by the dipper light. Once a head entry into the dipper was detected, there was a 2 second reward period to allow the rat to eat the pellet. The dipper light was then turned off, marking the end of the trial, and a 5 second intertrial interval (ITI) took place before the next trial commenced. ‘Incorrect’ responses (NP into an unilluminated port), ‘omissions’ (failure to respond during the limited hold), and ‘premature responses’ (NP during the ITI) were recorded and penalized with a 5 second timeout in which the house lights were turned off. A new trial resumed at the end of the timeout period and house lights were restored. ‘Perseverant responses’ (additional NP to the correct port following a correct response), and ‘timeout responses’ (NP during the timeout) were recorded but not penalized. ‘Latency to correct’ (time to correct), ‘latency to incorrect’ (time to incorrect), ‘latency to reward’ (time to collect reward), and ‘head entries’ (total NP in the food dipper) were also recorded. Rat chamber assignments were constant throughout the duration of all experiments and trial types were randomly selected such that each of the 5 stimulus ports was illuminated on 20% of trials.

Shaping

In order to reach optimal performance on this task, rats were first habituated to their individual testing chamber without reward for 3 daily sessions (5–10 min), and to the food pellets, which were available in their home cage. Next, rats were shaped in the operant boxes to collect a food reward on two daily sessions (20 min each). In these reward shaping sessions, a head entry to the food dipper triggered the release of one pellet and a 1 s illumination of the food dipper. This was followed by a 1 min ITI, after which a head entry triggered reward again. Next, in the stimulus shaping phase, the standard behavioral protocol was modified to have a 64 s stimulus duration, 2 s ITI, and 50 total trials. Over 2 months, stimulus duration was gradually decreased, and number of trials and ITI durations increased (see Table S1 for full schedule). During all phases, all rats received the same treatment so there was no individualized shaping.

Preoperative Training & Postoperative Testing

Once rats reached the target trial conditions (500 ms stimulus duration; 5 s ITI), they were trained on 15 sessions (T1-T15) of the standard task. Following surgery, rats were tested daily on the standard task for 9 days (D1-D9) and then presented with 4 challenge sessions (C1-C4), each one preceded by a standard session (S1-S4). Standard and challenge sessions were performed daily on alternate days (in the order: S1, C1, S2, C2, S3, C3, S4, C4).

Challenges 1 and 4: Reduced Stimulus Duration

Procedures for C1 and C4 were identical to the standard task except that stimulus durations were reduced to either 250 or 100 ms, respectively. The ability to attend to a shorter stimulus window and to respond appropriately to a stimulus that may be perceived as novel (or modified) was tested in this challenge. This is considered a top-down measure that is simply more challenging than the standard task. Rats with reduced goal-oriented capacity would display reduced performance in this task considering the stimulus is easier to miss.

Challenge 2: Distracting White Noise

The procedure for C2 was identical to the standard task except that a distracting burst of white noise (70 dB, 500 ms) was presented during the ITI. Noise onset occurred randomly at either 0.5, 2.5, 4.5, or 5 seconds into the ITI, such that each onset time occurred with equal frequency (25% of trials). Since the ITI directly precedes the stimulus presentation, this challenge tested the ability to stay on task in the presence of a task-irrelevant auditory distractor. Rats with a reduced capacity in top-down attention would display reduced performance because they would be overly distracted by the stimulus.

This challenge may also provide insight into bottom-up attention and/or the balance between bottom-up and top-down attention. This is because responding to a task-irrelevant stimulus, such as a loud noise that might signal danger, is evolutionarily advantageous. Therefore, healthy rats are expected to experience some reduction in performance in this challenge. A rat with exceptional top-down but reduced bottom-up capacity, on the other hand, might display “tunnel vision”. In this case, the inability to shift between top-down and bottom-up strategies potentially enhances performance in one task (in this case, the 5CSRTT) at the cost of another (survival, if the sound signaled danger). Therefore, rats with reduced bottom-up capacity may display enhanced performance in this task.

Challenge 3: Variable ITI

The procedure for C3 was identical to the standard task except that ITI durations varied randomly, lasting either 1, 3, 5, or 7 seconds with equal frequency. Since the onset time of the task-relevant stimulus varied, this challenge tested the ability to modify a learned goal-driven behavior. This is a top-down attention task because it increases the demand on vigilance, which is necessary for sustained attention and does not include novel bottom-up stimuli.

Behavioral Analysis

The 5CSRTT, which assesses goal-oriented behavior, is a task that primarily requires robust top-down attention. In the standard task, accuracy is considered the primary measure of top-down attention (Asinof and Paine, 2014). Omissions can also be a measure of attention deficit, but must be examined carefully to rule out the potential effects of reduced motivation or motility (Asinof and Paine, 2014). For instance, omissions in the presence of longer latencies to choice and/or reward, increased number of timeouts, and/or decreased number of head entries, would be considered a measure for motivation rather than attention. Latency to correct can be a measure for processing speed, although this must also be in the absence of measures that suggest reduced motivation or motility (Asinof and Paine, 2014). Finally, impulsivity can be measured by increased premature responses and compulsivity by perseverative errors (Asinof and Paine, 2014).

The primary measures we analyzed were percent accuracy (pAcc), percent omissions (pOmi), latency to correct (LatCorr), and percent correct (pCorr). For each session, the measures indicated number of correct trials divided by number of completed trials (pAcc), number of omitted trials divided by total number of trials presented (pOmi), time from onset of light stimulus to NP (LatCorr), and number of correct trials over total trials run, including omissions (pCorr). Additional measures were assessed and are reported in the supplementary materials (Figure S1). These include measures such as perseverative correct (pPers) which was calculated as the number of perseverant NP (NP into the correct port) divided by number of completed trials. Behavioral measures were collected by Med-PC (MedAssociates, Inc), outputted to a MedAssociates document, and extracted using a custom Matlab script (version R2017a, MathWorks).

Statistical Analysis

For the test phase, the 9 daily sessions were binned (3 sessions per bin), and called TestBin1, TestBin2, and TestBin3. BaseBin represented the last 3 training sessions before surgery and was used to compare pre- to postoperative performance. A repeated measures analyses of variance (rANOVA) was carried out for each behavioral measure separately using ‘group’ (lesion or sham) as the between-subject variable, and ‘bin’ and ‘session’ as the within-subject variables. Significant ‘group x bin’ differences were followed up by rANOVA where the BaseBin was compared to each TestBin individually. In addition, planned comparison of lesions versus shams for each bin and session were carried out using one-way ANOVAs. For each of the challenges, one-way ANOVAs compared the two groups (lesion vs sham). In addition, we compared each challenge (C) to the standard session (S) that preceded it (e.g., C2 vs S2). In these analyses, ‘group’ was the between-subject variable and ‘session type’ (standard vs challenge) was the within-subject variable. Finally, we also performed an overall rANOVA including all 8 sessions (4 challenges and 4 standard sessions), with ‘group’ as the between-subject variable, and ‘session type’ (standard vs challenge) and ‘challenge’ (C1–4) as within-subject variables. Statistical analyses were carried out in SPSS (version 24, IBM).

Results

Histology

The bilateral lesions of the dPPC with the newly delineated coordinates (Olsen and Witter, 2016), resulted in a mean percent damage in the left hemisphere of 70% (SD=16.4), with 92% spread (SD=12.2), and a mean percent damage in the right hemisphere of 61% (SD=10.1), with 86% spread (SD=13.9) (Figure 3). No rats displayed bilateral damage to the vPPC, a notable result considering that the delineated coordinates are meant to target the entire PPC (Olsen and Witter, 2016). This was likely a consequence of increased body weight compared to our pilot studies (unpublished), in which we reliably replicated the expected damage. Increased weight was unavoidable considering an ad lib diet and the extensive training required to reach pre-operative criterion on the 5CSRTT.

Figure 3 – Histology.

(A) Coronal sections showing representative areas (shown with shaded gray) as a result of NMDA lesions (n=10). (B) Percent observed cell loss for the right hemisphere (blue) and left hemisphere (orange) of lesioned animals (n=10) along the anterior-posterior axis. Solid lines are group medians and shaded areas display the range across maximum and minimum values for each hemisphere.

There was some damage to cortex areas surrounding the PPC, and damage was calculated as percent of analyzed sections that contained any bilateral damage to the same cortex area. Results showed two animals with 30–40% of sections displaying bilateral somatosensory damage, with the average damage across all subjects 24.1% +/− 9.6%. Five animals had 30–45% of sections with bilateral damage to the visual cortex, with average damage across all subjects at 28.9% ±10.6%. Only one animal showed damage in both somatosensory and visual cortices in greater than 30% of sections. All animals were checked for abnormal behavioral results. Due to a lack of outliers or significant differences between animals with greater versus lesser cortical damage, all animals were included in the behavioral results.

Performance in Standard 5CSRTT

Percent Accuracy

Rats with dPPC lesions displayed impaired percent accuracy in the standard 5CSRTT compared to controls in TestBin1 (Figure 4A). This was indicated by a ‘group x bin’ interaction (F_{3, 51} = 3.01, p = 0.039) in an overall rANOVA and followed up by individual rANOVAs showing that BaseBin was significantly different compared to TestBin1 (F_1,17 = 5.845, p = 0.027), but not compared to TestBin2 (p = 0.205) or TestBin3 (p = 0.356). In addition, planned comparisons (one-way ANOVA; lesion vs sham) revealed ‘group’ effects on D1 (F_1,18 = 5.506, p = 0.031), a trending effect on D2 (F_{1, 18} = 3.742, p = 0.070), and none on D3 (p = 0.478). Further analysis revealed a ‘session’ effect within TestBin1 (F_2,17 = 4.519, p = 0.018). Taken together, these results demonstrate that the lesion group displayed impaired top-down attention immediately after surgery in TestBin1 but not thereafter.

Figure 4 – Standard Task Results.

Results from the 5CSRTT task during test sessions, * denotes p <0.05. Daily sessions were binned (3 sessions per bin), and called TestBin1, TestBin2, and TestBin3. These were compared to baseline performance (Base), which consisted of the last three sessions before surgery. The lesion group initially displayed (A) lower accuracy (significant group x bin interaction), (B) lower percent correct (group x session significant omnibus ANOVA), and (C) longer latency to correct than sham rats (significant group difference in TestBin1). (D) In TestBin2, they displayed significantly more omissions (significant group difference).

Percent Correct

Consistent with the pAcc results, the lesion group also had reduced pCorr in TestBin1 (Figure 4B). This was demonstrated by a ‘group x session’ interaction in an overall rANOVA (F_2,34 = 3.439, p = 0.044) and group differences on D1 (F_1,18 = 5.206, p = 0.036) and D2 (F_1,18 = 5.140, p = 0.037) but not on any other testing or training days. This measure provides corroborating evidence that sustained top-down attention was impaired in TestBin1. Since pCorr takes omitted trials into account, these data indicate that increased omissions did not account for the observed group differences.

Latency to Correct

Rats with dPPC lesions demonstrated an increase in LatCorr in TestBin1 (Figure 4C). This was indicated by a ‘group’ effect in TestBin1 (F_1,18 = 6.991, p = 0.017), but not in TestBin2 (p = 0.194) or TestBin3 (p = 0.155), or in the BaseBin (p = 0.170). This means that rats in the lesion group were slower to make a correct choice in TestBin1 immediately following surgery.

While increased latency to correct could be interpreted as loss of motivation or locomotor ability (Asinof and Paine, 2014), this is unlikely considering that there was no evidence of simultaneous global inactivity (i.e., increased latency to reward, decreased number of head entries, decreased premature responses, and increased omissions). Instead, reduced latency to correct in TestBin1 could suggest a decrease in processing speed immediately after surgery.

Percent Omissions

Rats with dPPC lesions had higher percent omissions compared to controls in TestBin2 but not in TestBin1 (Figure 4D). This was indicated by one-way ANOVAs (lesion vs sham) showing ‘group’ effects in TestBin2 (F_1,18 = 4.818, p = 0.042), but not in TestBin1 (p = 0.565) or TestBin3 (p = 0.275). An rANOVA comparing base and test bins found no ‘group x bin” interaction (p = 0.288), though this was unsurprising considering an investigator error that disrupted our preoperative counterbalancing. As a consequence, group differences (lesion vs sham) in preoperative training were somewhat evident, although not significant (F_1,18 = 3.447, p = 0.081). Regardless, these results show that dPPC lesions resulted in increased omissions in TestBin2, which coincided with recovery from the impairments (decreased accuracy and increased latency to correct) observed in TestBin1.

Increased omissions can be interpreted in different ways. If there are concurrent decreases in global activity (i.e., longer latency to reward, longer latency to incorrect, less head entries, less premature responses), they are considered either locomotor impairments or motivation deficiencies (Asinof and Paine, 2014). In the absence of these global locomotor measures, however, omissions can be considered a measure of attention even when accuracy impairments are not observed (Asinof and Paine, 2014). In our case, increased omissions in TestBin2 were not accompanied by global inactivity measures and likely indicated impairment in the standard task, which measures top-down sustained attention. Considering the timing of observations (accuracy and latency to correct in TestBin 1, followed by omissions in TestBin2), these data suggest that the lesion group may have experienced a top-down impairment but employed shifting strategies throughout the progression of sessions. We explore this possibility further in the discussion.

Other measures

No other significant ‘group’ or ‘group x bin’ differences were found in any other measures, including latency to reward (Figure 1S–A), latency to incorrect, (Figure 1S–B), timeout responses (Figure 1S–C), premature responses (Figure 1S–D), or percent perseverative errors (Figure 1S–E). This indicates that dPPC lesions did not lead to loss of motivation or locomotor ability (latency to reward, latency to incorrect, timeouts, and head entries) and did not affect impulsivity (premature responses) or compulsivity (perseverative errors) (Asinof and Paine, 2014).

Performance in 5CSRTT Challenges

Global Performance

Across the challenge sequence as a whole, we compared all 4 challenge sessions to all 4 standard sessions, using ‘session type’ as the within-subject factor). No ‘group’ differences were observed in pAcc, pCorr or pOmi. In LatCorr, however, a ‘group’ effect (F_1,17 = 5.547, p = 0.031) indicated that rats with dPPC lesions had a global LatCorr impairment (Figure 5C). This means they were slower to make a correct choice, whether in a standard or challenge session. In the absence of other significant measures (i.e., longer latency to incorrect, longer latency to reward, less head entries, less premature responses), this likely indicates an overall impairment in processing speed, rather than an overall locomotor impairment (Asinof and Paine, 2014). Combined with the results from the standard task, this could be further evidence that the lesion group, although slightly impaired in top-down attention overall, was able to shift strategies. Considering rats were well-trained, it is reasonable to speculate that compensatory mechanisms, such as slowing down their choices, made up for any potential effects on accuracy.

Figure 5 – Challenge Tasks Results.

Results from the 5CSRTT challenges tasks; C1 250ms stimulus duration, C2 white noise distractor during cue presentation, C3 variable ITI between stimulus presentations, and C4 second shorter stimulus duration of 100ms. Each challenge was preceded by a standard session (S1, S2, S3, and S4). * denotes p <0.05. The lesion group displayed slower latencies to correct compared to control animals across all tasks, including the standard ones.

Reduced Stimulus Duration (C1 & C4)

Aside from the global latency impairments noted above, no significant group differences were observed in C1 or C4 specifically (Figure 5). As expected from a shortened stimulus challenge (Fizet, Cassel, Kelche, and Meunier, 2016), both groups performed worse in C1 and C4 compared to the standard task before it, as indicated by ‘session type’ effects in rANOVA. In C1 compared to S1, both groups had lower pAcc (F_1,17 = 30.485, p < 0.000), lower pCorr (F_1,17 = 26.574, p < 0.000), and higher LatCorr (F_1,17 = 6.218, p = 0.023). Similarly, in C4 compared to S4, both groups had lower pAcc (F_1,17 = 133.565, p < 0.000), lower pCorr (F_1,17 = 93.74, p < 0.000), and higher LatCorr (F_1,17 = 6.44, p = 0.021). No significant pOmi differences were observed compared to the standard task. These data confirm the effectiveness of the task and indicate that dPPC lesions did not impair sustained top-down attention in a reduced stimulus duration challenge.

Distracting White Noise (C2)

Similarly, there were no significant differences observed between groups when a distracting sound interrupted the ITI (Figure 5). The effectiveness of the challenge was indicated by a ‘session’ effect showing that both groups had higher pOmi (F_1,17 = 10.474, p = 0.005) in C2 compared to the standard session S2. There were no other ‘session type’ effects or ‘session x group’ interactions for pAcc, pCorr, or LatCorr. This indicates that, in general, subjects from both groups were distracted by the noise and reacted to it by omitting trials rather than guessing incorrectly. Our results suggests that dPPC lesions did not impair selective attention for a task-relevant stimulus over a task-irrelevant one compared to controls. Since this challenge also requires attentional balancing between top-down and bottom-up strategies, and since both groups had similar increases in omissions, the results also suggest that bottom-up attention (the ability to be distracted) is preserved in animals with dPPC lesions. In other words, the dPPC lesion did not prevent an adaptive switch from top-down to bottom-up strategy in response to the loud noise (which could potentially signal danger).

Variable ITI (C3)

We observed that the lesion group took longer to make a correct choice when the stimulus was presented at varying intervals, although this occurred both in C3 and the preceding standard session S3. This was indicated by a main effect of group in LatCorr (F_1,17 = 5.741, p = 0.028), with no interaction of ‘session x group’. No other group or ‘session x group’ differences were observed for pAcc, pCorr, or pOmi. The effectiveness of the challenge (i.e., increased difficulty level) was indicated by increased omissions for all subjects (F_1,17 = 24.295, p < 0.000) in C3 compared to S3. No differences were observed in pAcc or LatCorr between C3 and S3. Results were similar across the 4 possible ITI intervals (1, 3, 5, and 7 sec). This demonstrates that the challenge was effective in increasing difficulty of the task and that all rats responded by omitting trials rather than guessing incorrectly. The lesion group took longer to respond during correct trials, but since this was true for both the standard and challenge session, it is likely related to the global LatCorr increase mentioned above.

Since the ability to respond successfully to short ITI trials (i.e., 1 and 3 sec) requires rats to be ‘distracted’ by the unexpected stimulus, these results could provide further evidence that bottom-up attention was intact in rats with dPPC lesions. Although the lesion group was able to maintain stable performance across multiple trials once they understood the ‘rules’ (in terms of accuracy and omissions), increased choice time for correct trials suggests a deficiency in processing speed, which is consistent with top-down disruptions in the standard task.

Discussion

Rats with neurotoxic lesions to the dPPC displayed impairments in sustained top-down attention (decreased accuracy in TestBin1; increased omissions in TestBin2) and processing speed (increased latency to correct in TestBin1), but recovered by the end of the standard 5CSRTT. In the challenge tasks, processing speed appeared to be globally affected (increased latency to correct) but general performance on individual tasks (accuracy and omissions) was preserved. No evidence of impulsivity (premature responses) or compulsivity (perseverative errors) was observed, and the lesion group was not disrupted in tasks that rely on a balance between top-down and bottom-up strategies (distracting noise and ITI challenges). Selective attention was also unaffected in the lesion group (distracting noise challenge). Taking all behavioral measures into account, these results indicate that dPPC lesions caused a slight impairment in top-down attentional processing. They also suggest that bottom-up processing was not affected.

The timing of the various impairments further suggests that compensatory mechanisms were likely at play. Task performance (accuracy and omissions) was initially impaired but then recovered within a few sessions and was later followed by slower decision-making (increased latency to choice) in the challenges. One possibility is that after having underperformed in the standard task, the lesion group altered their strategy when the task rules became more varied and difficult. This compensatory behavior could indicate underlying neural compensation, for example, in the form of slower integration of sensory stimuli in order to subserve a well-learned rule. Overall, our results suggest that although intact rodent dPPC supports top-down attention, performance levels can be recovered at the cost of slower processing speed. Our results are consistent with the role of IPL in humans and monkeys (Behrmann, Geng, and Shomstein, 2004; Shomstein, 2012), and are in line with the hypothesis that intact rodent dPPC and vPPC have functionally dissociated roles in sustained visual attention. Whether or not there is direct evidence for this functional dissociation will be the subject of future studies.

The PPC is involved in sustained visuospatial attentional processing

Contrary to previous studies, our results demonstrate that the dPPC supports sustained attention, as measured by the 5CSRTT. In a previous study, PPC lesions made by inactivation of cholinergic afferents did not result in 5CSRTT impairments (Maddux et al., 2007). Similarly, another study found that broad neurotoxic lesions to the PPC also did not result in any impairments (Muir et al., 1996). The discrepancy with our results could reflect methodological differences such as updated anatomical coordinates (Olsen and Witter, 2016), specific targeting of the dPPC, and lesion strategy (neurotoxic lesions versus cholinergic inhibition). Alternatively, it could be an analysis discrepancy. In Maddux et al, it is important to note that analyses focused on comparing continuous versus partial reinforcement. A visual representation of the percent correct measure appears similar to ours in the first 5 days post-treatment, although no direct statistical comparisons were presented between the PPC and control groups. It is possible that an a priori re-analysis of their data that excludes reinforcement as an independent variable reproduces our results. This may also be true for percent accuracy and omissions since these raw data were not provided. Thus, our data may not necessarily contradict previous studies. Finally, since the sham group underwent only craniotomies, it is possible that the observed effects reflected general deficiencies resulting from the surgery itself. Rather than a mild specific deficit in sustained attention, the results could indicate different sensitivities across measures to a broader but temporary deficit, such as greater generalization across stimuli or spatial locations. However, considering that the lesioned group was fully spared in most measures, we find this interpretation of global impairment unlikely.

Our results do fall in line with electrophysiological studies showing that PPC cells are recruited during sustained visuospatial attention tasks. Two studies make use of a novel task that is adapted from the 5CSRTT, where the monitored locations are projected on a Floor Projection Maze (Furtak, Cho, Kerr, Barredo, Alleyne, Patterson, and Burwell, 2009) instead of wall ports. In the first study, results show that PPC cells activate in response to several task epochs, including onset of the illumination signal and selection of the correct location, and that PPC cells were phase locked to the theta frequency (Yang et al., 2017). In another study, vPPC neurons were found to respond to the illumination signal faster than dPPC neurons, suggesting a functional differentiation between these areas (Yang et al., 2021). These studies are consistent with another showing that PPC cells are involved in signal detection during a similar sustained attention task (Broussard et al., 2006). In light of evidence from these studies, our results support the hypothesis that the intact PPC is involved in sustained visuospatial attention.

dPPC lesions impaired top-down attention

The 5CSRTT, which assesses goal-oriented behavior, is a task that primarily requires robust top-down attention. In this study, a lesion to the dPPC significantly reduced both accuracy and latency to correct in the first 3 post-operative trials (TestBin1). These results occurred in the absence of measures that could collectively indicate lack of motivation or reduced motility. In the following trials (TestBin2), the impairment shifted, so while accuracy and latency to correct recovered, omissions increased. In the absence of measures that could collectively indicate reduced motivation/mobility, this suggests that attentional processing was still compromised. Finally, during the challenge trials, lesioned rats displayed globally reduced latency to correct in the absence of any other effects. This suggests that the rats adopted a more deliberate strategy, taking longer to make a decision in order to preserve performance rates. Taken together, our results indicate that dPPC lesions impaired sustained top-down attention and suggest that processing speed was globally impaired, potentially as a compensatory mechanism involving more deliberate decision-making.

dPPC lesions may not have affected bottom-up attention

Although mostly a top-down task, the 5CRSTT can also be used to assess bottom-up processing to some extent, or at least the ability to suppress responses that are not goal-driven (i.e., impulsivity) and the ability to balance task-relevant with task-irrelevant stimuli (i.e., selective attention). In terms of impulsivity, we did not observe any premature responses (i.e., responses during the inter-trial interval) across any of the standard sessions (TestBin1-3), or in the variable ITI challenge (C3). This indicates that suppression of bottom-up impulses was appropriate in spite of the lesions, which indicates that dPPC is not critical for general impulse control. In terms of selective attention, it appears that dPPC lesions did not impair the ability to selectively attend to a relevant stimulus over an irrelevant one in the white noise challenge (C2). On the other hand, lesions did not lead to enhanced performance in the white noise challenge either. This might be expected from rats that are hyper-focused (a disadvantage from a survival stand point) and unable to direct attention to the task-irrelevant stimulus. Since the lesion group was able to shift between top-down and bottom-up strategies and was momentarily distracted by the noise, our results may lend further evidence to the hypothesis that the dPPC is not critical for bottom-up strategies in the intact rat brain.

Finally, another measure that may support the idea that bottom-up processing was not impaired, is accuracy for shorter-than-expected ITIs in C3. Presumably, the ability to be ‘distracted’ by a stimulus that is task-relevant but temporally unexpected helps intact rats perform this task. Deficient bottom-up processing may lead to decreased accuracy because it takes rats longer to realize that the ITI is shorter. In our study, however, dPPC lesions did not decrease accuracy in this challenge, suggesting lesioned rats benefitted from the temporal ‘distraction’ equally compared to controls. Although this is not a validated measure for bottom-up processing, it could support the hypothesis that dPPC does not regulate bottom-up processing, especially when taken together with the absence of impulsivity, and intact selective attention. Further studies will need to be performed to fully elucidate the role of dPPC in bottom-up attention.

Performance deficits were rescued at a cost to processing speed

Although our results demonstrate that dPPC lesions initially impaired sustained top-down attention, performance levels quickly recovered within a few sessions. This likely reflected behavioral compensation, especially considering the trajectory of observed impairments, starting with percent accuracy and latency to correct in TestBin1, followed by omissions in TestBin2, and finally by an overall increase in latency to correct. Rats may have probed different behavioral strategies to make up for their initial accuracy impairment, first by cautiously omitting trials rather than guessing, and in general by taking longer to make a decision. This putative deliberate behavior suggests that processing speed may have been reduced in the service of maintaining a steady level of performance. In other words, although the overall drive to adhere to a goal was preserved, as well as the goal itself, the dPPC lesions decreased the ability to apply the known task rules, perhaps by reducing the capacity to integrate appropriate stimuli and use them to direct goal-oriented attention. Alternatively, increased latencies to correct could be related to post-error slowing (Li, Huang, Yan, Paliwal, Constable, and Sinha, 2008) or interpreted as maladaptive top-down processing. In either case, investigating the neural underpinnings of this compensation will help to elucidate the role of the PPC in attention, and the potential dissociation between dPPC and vPPC.

Behavioral strategies were likely accompanied by underlying neural compensation

dPPC lesions likely resulted in neural compensation to support optimal performance in the 5CSRTT. This would be unsurprising considering that attention is a distributed process involving diverse cell populations throughout the frontoparietal network and other neocortical areas (Katsuki and Constantinidis, 2014). In addition, as a multimodal association area, the PPC is vastly interconnected to multiple regions that could be candidates for compensatory neuroplasticity (Broussard, 2012; Oh, Son, Morris, Choi, Lee, and Rah, 2021). For example, the PPC is strongly interconnected with thalamic, sensory (visual, auditory, and somatosensory), and motor areas where neurons are tuned to a variety of stimulus-specific and spatial elements (Bizley and Walker, 2010; Mesulam, 1998; Priebe and Ferster, 2008). Compensatory neuroplasticity could occur via distribution of cognitive load across other cortical areas and/or through novel neuronal pathways, both of which have been widely observed across species, functional networks, disorders, and developmental timepoints (Alaverdashvili and Whishaw, 2010; Beauparlant, van den Brand, Barraud, Friedli, Musienko, Dietz, and Courtine, 2013; Casella, Thomas, Vanino, Fellows-Mayle, Lifshitz, Card, and Adelson, 2014; Fuchs and Flugge, 2014; Grafman and Litvan, 1999; Park, Lee, Kim, Lee, Lee, and Oh, 2010; Rauschecker, 2002; Robbins, 1992; Voss, Thomas, Cisneros-Franco, and de Villers-Sidani, 2017; Waggener and Coppola, 2007). Alternatively, multiple competing networks could undergo rearrangement whereby a non-dominant network becomes the dominant one following a lesion. This has been observed, for instance, in memory processes, including encoding and retrieval of contextual fear memories (Kanishka and Jha, 2023; Mysore and Kothari, 2020).

Role of acetylcholine in dPPC-mediated sustained attention

Interestingly, the fact that PPC neurotoxic lesions in our study impaired attention but cholinergic inactivation did not (Maddux et al., 2007) may speak to the neural underpinnings of attention control. Although several lines of evidence demonstrate that acetylcholine is necessary for attentional processes in certain areas, including prefrontal cortex (Dalley, Theobald, Bouger, Chudasama, Cardinal, and Robbins, 2004; Gill, Sarter, and Givens, 2000; Thiele and Bellgrove, 2018), it could be that dPPC outputs regulate sustained attention via other neuromodulators such as norepinephrine, which is also known to support sustained attention (Smith and Nutt, 1996; Thiele and Bellgrove, 2018). Alternatively, it could be that lesion strategy does not account for the differences in our results compared to Maddux et al. (2007), in which case our results may not inform the neuromodulation underlying sustained attention.

PPC coordinates are age/weight specific

We note that the coordinates used in this study have previously been reported to target the entire PPC, not only dPPC. In fact, we performed pilot surgeries (data not shown) that resulted in complete ablation of the PPC, as predicted. However, both the previous studies and our pilot surgeries were performed in young rats (~2 months) with an average weight of (250–350 g). In the current study, by the time rats were fully trained in the 5CSRTT and ready for surgery, they were older (~5 months) and heavier (400–460 g). This likely led to a shift in anatomical coordinates, a notable point for future studies to consider when targeting PPC.

Future directions

In order to assess a possible dissociation between top-down and bottom-up attentional processing in the PPC, it will be necessary to compare lesions to either the dPPC or vPPC, possibly using transient, reversible inactivation by means of designer receptors exclusively activated by designer drugs (DREADDs). The flexibility afforded by this technique could result in longer-lasting performance impairments that are unmasked by compensatory effects. In addition, it would be beneficial to incorporate tasks that measure bottom-up attention more directly. For example, the PPC could be lesioned before training to evaluate acquisition as well maintenance of the task. In theory, acquisition could be more susceptible to bottom-up disturbances considering this is a period when rats are trying to assign value to unknown stimuli and testing contingencies. Re-running the standard training using a different task-relevant stimulus, possibly in a different sensory modality, might also be informative. Finally, in order to increase translational validity, future studies should include both male and female rodents.

Conclusion

In this study, we demonstrated that neurotoxic lesions to the dPPC result in progressive impairments that shift in nature across the experimental timeline. In the first three days of the standard task, accuracy impairments and longer latencies to correct in the lesion group are evident. In the following days, accuracy and latency stabilize but omissions increase briefly. Across a series of challenges that follow, latency to correct increases globally across all tasks in the lesion group, suggesting overall reduced processing speed and/or maladaptive post-error slowing. These results demonstrate behavioral compensations that may indicate underlying compensatory neuroplasticity. Our primary interpretation of these results is that top-down attentional processes were compromised whereas bottom-up processes may have been intact.

Our results are in line with evidence from humans and non-human primates showing a functional dissociation between the dPPC and vPPC. Although we know that this dissociation exists in primates, it is not entirely clear how attention shifts between top-down and bottom-up strategies to support higher order cognitive function. Rodent models will be key in this ongoing investigation of attentional circuits, so it is critical to establish more directly whether the functional dissociation exists in rodents.

Highlights.

• Rats with dPPC lesions showed transient accuracy decreases in an attention task

• Recovery in accuracy and latency was accompanied by an increase in omissions

• In challenge probe tests, lesioned rats showed elevated latency to correct

• Results implicate the dPPC in top-down attention, specifically processing speed