The Dynamic Multisensory Engram: Neural Circuitry Underlying Crossmodal Object Recognition in Rats Changes with the Nature of Object Experience

Abstract

Rats, humans, and monkeys demonstrate robust crossmodal object recognition (CMOR), identifying objects across sensory modalities. We have shown that rats' performance of a spontaneous tactile-to-visual CMOR task requires functional integration of perirhinal (PRh) and posterior parietal (PPC) cortices, which seemingly provide visual and tactile object feature processing, respectively. However, research with primates has suggested that PRh is sufficient for multisensory object representation. We tested this hypothesis in rats using a modification of the CMOR task in which multimodal preexposure to the to-be-remembered objects significantly facilitates performance. In the original CMOR task, with no preexposure, reversible lesions of PRh or PPC produced patterns of impairment consistent with modality-specific contributions. Conversely, in the CMOR task with preexposure, PPC lesions had no effect, whereas PRh involvement was robust, proving necessary for phases of the task that did not require PRh activity when rats did not have preexposure; this pattern was supported by results from c-fos imaging. We suggest that multimodal preexposure alters the circuitry responsible for object recognition, in this case obviating the need for PPC contributions and expanding PRh involvement, consistent with the polymodal nature of PRh connections and results from primates indicating a key role for PRh in multisensory object representation. These findings have significant implications for our understanding of multisensory information processing, suggesting that the nature of an individual's past experience with an object strongly determines the brain circuitry involved in representing that object's multisensory features in memory.

Significance statement: The ability to integrate information from multiple sensory modalities is crucial to the survival of organisms living in complex environments. Appropriate responses to behaviorally relevant objects are informed by integration of multisensory object features. We used crossmodal object recognition tasks in rats to study the neurobiological basis of multisensory object representation. When rats had no prior exposure to the to-be-remembered objects, the spontaneous ability to recognize objects across sensory modalities relied on functional interaction between multiple cortical regions. However, prior multisensory exploration of the task-relevant objects remapped cortical contributions, negating the involvement of one region and significantly expanding the role of another. This finding emphasizes the dynamic nature of cortical representation of objects in relation to past experience.

Keywords: c-fos; cross-modal; multisensory integration; parietal cortex; perirhinal cortex.

Figure 1.

Schematic illustrating procedural details for PRh (Experiments 1–3) and PPC (Experiments 4 and 5) intracranial administration experiments. In Experiments 1a and 4a, lidocaine or saline was infused (downward arrows) directly into PRh or PPC, respectively, immediately before the sample phase in the tactile-only (A), visual-only (B), and crossmodal (C) object recognition tasks. In Experiments 1b and 4b, lidocaine or saline was infused into PRh or PPC, respectively, immediately before the choice phase in the tactile-only (A), visual-only (B), and crossmodal (C) object recognition tasks. All tasks were run in Experiments 1 and 4 with a retention delay of 1 h between sample and choice phases. For the PE/CMOR experiments (D), lidocaine or saline was administered into PRh or PPC (Experiments 2 and 5, respectively) immediately before (Experiments 2a and 5a) or after (Experiments 2b and 5b) the multimodal preexposure session or immediately before the sample (Experiments 2c and 5c) or choice (Experiments 2d and 5d) phase. There was a 24 h delay between preexposure and sample in all PE/CMOR experiments and a 3 h retention delay between sample and choice phases in Experiments 2 and 5. Experiment 3 was identical to Experiment 2c, except that the retention delay between sample and choice was 1 h.

Figure 2.

Cannula placements in PRh and PPC. A, Placements of infusion cannula tips in PRh from rats in a representative experiment (Experiment 2b; n = 12). All rats included in behavioral analyses had cannulas located between 5.88 and 6.12 mm posterior to bregma and near the border between areas 35 and 36. B, Microphotograph (1.5× magnification) showing a guide cannula tract in the left hemisphere of a rat from Experiment 2b; the arrow indicates the infusion cannula tip. C, Placements of infusion cannula tips in PPC from rats in a representative experiment (Experiment 4a; n = 12). All rats included in behavioral analyses had cannulas located between 3.24 and 3.84 mm posterior to bregma. D, Microphotograph (1.5× magnification) showing a guide cannula tract in the left hemisphere of a rat from Experiment 4a; the arrow denotes the infusion cannula tip.

Figure 3.

Object recognition performance of rats administered saline or lidocaine into PRh immediately before the sample phase of the CMOR, tactile, and visual SOR tasks (n = 16; A) or immediately before the choice phase of the CMOR, tactile, and visual SOR tasks (n = 12; B). Retention delay in all conditions was 1 h. Data are presented as the average discrimination ratio (±SEM). *p < 0.05; **p < 0.01, saline versus lidocaine.

Figure 4.

Object recognition performance in the PE/CMOR task by rats receiving infusions of saline or lidocaine into PRh immediately before the preexposure phase (n = 13; A), immediately after the preexposure phase (n = 12; B); immediately before the sample phase (n = 9; C), or immediately before the choice phase (n = 12; D). For all experiments, there was a 3 h retention delay between the sample and choice phases; with this delay length, normal rats typically fail to show novelty preference in the CMOR task without multimodal preexposure. Data are presented as the average discrimination ratio (±SEM). *p < 0.05; **p < 0.01, saline versus lidocaine.

Figure 5.

Object recognition performance of rats receiving infusions of saline or lidocaine into PRh immediately before the sample phase of the PE/CMOR task with a 1 h retention delay between the sample and choice phases (n = 10). This experiment controls for the difference in retention delays between CMOR (Experiment 1) and PE/CMOR (Experiment 2) tasks. Data are presented as the average discrimination ratio (±SEM). **p < 0.01, saline versus lidocaine.

Figure 6.

Object recognition performance of rats administered saline or lidocaine into PPC immediately before the sample phase of the CMOR, tactile, and visual SOR tasks (n = 12; A) or immediately before the choice phase of the CMOR, tactile, and visual SOR tasks (n = 10; B). Retention delay in all conditions was 1 h. Data are presented as the average discrimination ratio (±SEM). *p < 0.05; **p < 0.01; ***p < 0.001, saline versus lidocaine.

Figure 7.

Object recognition performance in the PE/CMOR task by rats receiving infusions of saline or lidocaine into PPC immediately before the preexposure phase (n = 11; A), immediately after the preexposure phase (n = 9; B), immediately before the sample phase (n = 11; C), or immediately before the choice phase (n = 7; D). For all experiments, there was a 3 h retention delay between the sample and choice phases with this delay length. Normal rats typically fail to show novelty preference in the CMOR task without multimodal preexposure. Data are presented as the average discrimination ratio (±SEM).

Figure 8.

A, Fos expression levels in PRh, PPC, primary auditory (Aud), and motor (M1 and M2) cortex from rats exposed to 30 novel objects (visual and tactile exploration allowed; preexposure, black bars; n = 8) compared with rats exposed to the empty apparatus for the same amount of time (“Empty,” white bars; n = 8) 1 h before being killed. Significant upregulation was observed only in PRh. Normalized counts of Fos-positive cells are presented as mean ± SEM. B, C, Microphotographs (5× magnification) of representative slices illustrating Fos-immunoreactivity in PRh (∼5.2 mm posterior to bregma) from a rat in the empty (B) and preexposure (C) conditions; note the substantially greater number of darkly stained cells in C. **p < 0.01. RS, Rhinal sulcus. Scale bar, 200 μm.

Figure 9.

Topographic Fos expression levels in PRh areas 35 (A) and 36 (B) from rats exposed to 30 novel objects (visual and tactile exploration allowed; preexposure, black bars; n = 8) compared with rats exposed to the empty apparatus for the same amount of time (“Empty,” white bars; n = 8) 1 h before being killed. Significant upregulation was observed in the middle portions of PRh area 35, whereas area 36 displayed the greatest increases in the rostral and caudal regions. Normalized counts of Fos-positive cells are presented as mean ± SEM. **p < 0.01.

Figure 10.

Topographic Fos expression levels in PRh areas 35 (A) and 36 (B) of rats in various combinations of preexposure and sample conditions: no preexposure/no sample (n = 4); preexposure/no sample (n = 4); no preexposure/sample (n = 4); and preexposure/sample (n = 4). Rats were killed 1 h after the second phase (sample or no sample). Rats that experienced multimodal object preexposure followed 24 h later by a tactile sample phase (reexposure to the same objects) had significantly higher levels of Fos expression throughout all levels of PRh. B, C, Microphotographs (10× magnification) of representative slices illustrating Fos-immunoreactivity in PRh (∼5.2 mm posterior to bregma) from a rat in no preexposure/sample (C) and preexposure/sample (D) conditions; note the greater number of darkly stained cells in D. **p < 0.01; ***p < 0.001. RS, Rhinal sulcus. Scale bar, 200 μm.