The contribution of cross-modal sensory input to reaching and grasping in blind and sighted short-tailed opossums

Abstract

Reaching and grasping are essential goal-directed behaviors that require the integration of visual, olfactory, and somatosensory inputs. Although the loss of vision can profoundly disrupt sensory-guided behaviors, mammals often exhibit compensatory cross-modal plasticity allowing them to reach and grasp with ease. To examine how early sensory loss shapes reaching behavior, we performed bilateral enucleations in short-tailed opossums (Monodelphis domestica) at postnatal day 4, before retino-thalamic and thalamocortical connections have formed. We assessed performance in early blind (EB) and sighted control (SC) animals using a semi-naturalistic reach-to-grasp task requiring precise unilateral limb targeting to retrieve a dead cricket. To isolate the contributions of specific sensory modalities to this task, we selectively disrupted olfactory and mystacial vibrissae inputs and manipulated lighting conditions during task performance. EB opossums were capable of accurate reaching and grasping, although SC animals outperformed EB opossums under light conditions, but not in the absence of light. Both groups relied strongly on tactile input, as whisker trimming significantly increased targeting error. Removal of olfactory input also impaired performance, with a disproportionately greater effect in EB animals. These findings demonstrate that short-tailed opossums retain functional reach-to-grasp behavior after early vision loss and that accurate forelimb movements are generated by the enhancement of the spared sensory systems.

Introduction

Reaching and grasping is an ethologically relevant behavior for many animals and frequently demands complex multimodal sensory processing and sensorimotor integration. While most studies that examine these behaviors are performed in non-human primates, reaching and grasping is a ubiquitous behavior for most mammals with well-developed forelimbs and forepaws. Although scientists often focus on understanding how a single sensory system contributes to species-specific behaviors, in natural environments animals rely on the integration of multiple sensory modalities in a context-dependent manner to generate adaptive behavior. Effective motor control likely depends on the integration of visual, auditory, proprioceptive, tactile, vestibular and even olfactory inputs that contribute to different stages of movement planning and execution (Stone and Gonzalez, 2015).

Because of the importance of multimodal processing, loss of one sensory modality, whether congenital or acquired, profoundly alters sensory experience and can severely disrupt normal behaviors. Even when faced with such disruptions, mammals exhibit remarkable flexibility, compensating both neurophysiologically and behaviorally, particularly when sensory loss occurs during early development (Kahn and Krubitzer, 2002; Karlen et al., 2006; Fujii et al., 2009; Wong et al., 2011; Englund et al., 2020; Ramamurthy et al., 2021; Asumbisa et al., 2022). Over the past four decades, studies in humans, non-human primates, ferrets, mice, rats, and hamsters have shown that sensory loss can trigger large-scale cortical reorganization and, in some cases, enhancements of remaining sensory modalities (Sharma et al., 2000; Izraeli et al., 2002; Chabot et al., 2008; Voller et al., 2014; Hagan et al., 2017; Wang et al., 2017; Maruoka et al., 2024). However, most of these studies examined sensory loss at relatively late stages of development when anatomical pathways are partially or completely developed and only examined the loss of a single sensory modality.

In the current study, we examine the effects of early loss of vision on reaching behavior in the short-tailed opossum (Monodelphis domestica), and the contribution of tactile and olfactory inputs on task performance. The short-tailed opossum is precocial, allowing us to remove visual inputs via bilateral enucleations before retino-thalamic and thalamo-cortical connections have formed (postnatal day 4; P4). Previous studies in our laboratory demonstrate that these early enucleations lead to substantial changes in the connectivity and response properties of neurons in the putative visual cortex and somatosensory cortex (Kahn and Krubitzer, 2002; Karlen et al., 2006; Ramamurthy and Krubitzer, 2018). For example, neurons in the primary visual cortex (V1) in early blind (EB) short-tailed opossums respond to auditory and tactile stimulation (Figure 1A), and neurons in the whisker representation in primary somatosensory cortex (S1) undergo receptive field modifications that may subserve the greater texture discrimination abilities observed in blind animals (Kahn and Krubitzer, 2002; Ramamurthy and Krubitzer, 2018; Ramamurthy et al., 2021; Figure 1B,C). Interestingly, EB opossums also outperform SC opossums on tasks where tactile input from the whiskers compensates for the lack of vision (Englund et al., 2020; Ramamurthy et al., 2021).

Figure 1: Alterations in the functional organization of V1 and receptive fields for neurons in V1 and S1 following early loss of vision.

(A) The organization of neocortex in sighted (left) and early blind opossums (right). The neurons in re-organized V1 and other adjacent visual areas of early blind opossums respond to auditory and tactile stimulation. (B) Bar graph showing the proportion of neurons in re-organized V1 and other “visual areas” with receptive fields representing the face, head, and lower extremities of early blind short-tailed opossums. (C) Line graph showing the reduction in receptive field size of neurons in the whisker representation of S1 in early blind (red) and sighted control (blue) short-tailed opossums. Adapted from Khan et al, 2002 (A and B) and Ramamurthy et al, 2019 (C).

We assessed the extent to which EB short-tailed opossums compensate for lack of vision using their remaining sensory modalities by assessing their performance on a semi-naturalistic and widely used reach-to-grasp task that requires precise sensory-guided targeting of the forelimb and paw to retrieve a reward (Ivanco et al., 1996; Whishaw, 1996; Klein et al., 2012; Whishaw et al., 2017). To determine the contribution of individual sensory modalities to reach targeting after vision loss, we performed a series of experiments in which we selectively and reversibly removed input from individual sensory modalities. These manipulations included extinguishing light sources (for sighted controls), trimming mystacial vibrissae, and temporarily inducing anosmia. To our knowledge, this is the first investigation to examine the contribution of multiple sensory inputs to this ethologically important behavior.

Methods

Reaching behavior was examined in 21 adult (>180 days old) short-tailed opossums (Monodelphis domestica) weighing 70 to 120 grams. Eleven animals (7 females, 4 males) were bilaterally enucleated (EB) at postnatal day 4 (P4), and 10 animals (7 females, 3 males) served as sighted littermate controls (SC). All animals were born and housed in the UC Davis Psychology Department vivarium under a 14:10 hour light/dark cycle (lights on at 6 AM), with food and water available ad libitum. Animals were weaned at 56 days and separated from littermates at four months. All procedures were approved by the UC Davis Institutional Animal Care and Use Committee and followed NIH guidelines.

Bilateral Enucleation Surgery:

Bilateral enucleations were performed at P4 as described previously (Ramamurthy et al, 2018). Briefly, dams were anesthetized with isoflurane (5%) and maintained under Alfaxalone (3.0 mg/kg IM). Pups were anesthetized using hypothermia. The skin overlying the developing eye was incised, the immature retina and optic tissue removed, and the socket flushed with sterile saline (9%). The incision was sealed with surgical glue (GLUture, Zoetis Inc. Kalamazoo MI). Half the litter was enucleated, while the remaining pups served as sighted controls. Dams recovered in their home cage following IACUC guidelines.

Behavioral Testing:

Adult EB and SC opossums were trained for five days to perform a reach-to-grasp task. The reaching box was modeled based on that described previously for rodents (Whishaw et al., 2010; Alaverdashvili and Whishaw, 2013; Figure 2A). The box was made of clear Plexiglass (350 × 250 × 300 mm). A vertical slit centered in the front of the box (10 mm wide, 150 mm tall) and served as an opening through which opossums could reach for a dead cricket reward placed on a shelf attached to the outside of the reaching box. This shelf was positioned 40 mm above the floor of the box, and the cricket was placed on a well in the shelf 25 mm away from the slit opening (Figure 2B).

Figure 2: Reaching and grasping test box and testing paradigm for the short-tailed opossum.

(A) Front view of the reaching and grasping box constructed for short-tailed opossums. Animals were trained to reach through a slit for a cricket reward placed on a platform approximately 25mm away from the slit opening. The cricket reward was positioned in a central well to ensure it remained in place. The dimensions of the box are shown in the schematic. (B) Image depicting the size and shape of the cricket reward relative to well size (circle around cricket. (C) Animal training and testing paradigm. Animals were shaped for five days and were tested after achieving a weighted error below 40%. Once criterion performance was achieved, animals were tested under light and dark conditions with whiskers intact (baseline), whisker-trimmed, and anosmic. To minimize potential order effects associated with task learning, the sequence of baseline and whisker trimmed testing, as well as light and dark testing conditions, was counterbalanced across animals. (D) Schematic illustrations depicting the four reach outcome categories scored during the task: Successful, Swipe, Contact, and Miss.

Training consisted of placing short-tailed opossums inside the reaching box for 10-minute sessions, during which dead crickets were placed around the slit and on the platform. All animals underwent a standardized shaping procedure to learn the association between the slit, platform and reward, although some animals more readily initiated reaching behavior than others. During shaping, crickets were initially positioned near or within the slit and were gradually moved to the target indentation on the platform as animals learned to associate the slit with reward delivery. As learning progressed, cricket placements were restricted to the platform, and the distance from the slit was increased to shape reaching behavior. To discourage repeated reaching attempts for a single reward, which were common, animals were limited to one attempt per trial by alternating placement of the target cricket on the platform with a second reward placed at the rear of the testing chamber. Animals were considered shaped once they reliably reached for crickets placed on the target indentation, regardless of success rate. Following shaping, animals underwent 5 days of training, and performance was evaluated using a weighted error metric. Animals that did not reach the performance criterion below 40% weighted error within 5 days were excluded from further testing.

Reach-to-grasp tests consisted of 20–30 trials per session in which dead crickets were presented over a maximum period of 10 minutes. Cricket size ranged from 5–10 mm in length (Figure 2B). Data collection began after 5 days of training. Each Visual Group (EB, SC) was tested under three Sensory Conditions: baseline, whisker-trimmed, and anosmic. The order of baseline and whisker-trimmed testing was counterbalanced across animals. Testing was also conducted under both light and dark conditions (Lighting Condition), which were counterbalanced across consecutive days (Figure 2C). A 660 nm wavelength red lamp was used to provide visibility for experimenters during trials conducted in darkness. Previous studies from our laboratory show that performance under this illumination does not differ from complete darkness (Englund et al., 2020), and it is known that short-tailed opossums possesses long-wavelength sensitive (LWS) cones with an expected peak sensitivity in the ~500 – 570 range (Hunt et al., 2009), placing 600 nm well outside this range. However, this does not completely rule out residual photoreceptor activation under these conditions.

Whisker trimming and induction of anosmia in adults:

To determine the extent to which other sensory systems contributed to task performance, we first removed facial whiskers and then induced anosmia after allowing whiskers to regrow. For whisker trimming, animals were briefly anesthetized with isoflurane (2–5%), and all mystacial, submandibular, and genal vibrissae were trimmed to ~1 mm in length. Following whisker trimming, behavioral testing was done for two consecutive days. Whiskers were allowed to grow back (~30 days), and then anosmia was induced by intranasal infusions of zinc sulfate (50 μL, 5%) into each nostril, administered 1 hour apart. We generated sham controls in a subset of animals (5 blind, 7 control), in which sterile saline (50 μL, 9%) was infused into each nostril under the same anesthetic regime and at the same time intervals stated above. Animals were allowed to recover inside their home cage for 10–12 hours before undergoing sham or anosmia reach testing.

In animals that had undergone sham and zinc sulfate infusions, we conducted an olfactory habituation/dishabituation experiment to ensure that anosmia was successfully induced. The task consisted of four vehicle presentation epochs (Vehicle 1 – 4; mineral oil; Thermo Fisher Scientific, Waltham, Massachusetts) followed by four odor exposure epochs (Odor 1 – 4). Each epoch lasted 4 minutes and was separated by a 1-minute intertrial period. The odor of choice was varied across anosmic and sham testing to maximize the novelty effect of each odor. Odors used were hexyl acetate, hexanoic acid, amyl acetate, and ethyl acetate (all 1%, Thermo Fisher Scientific, Waltham, Massachusetts), each diluted in mineral oil.

Three independent observers blind to the animals’ anosmia condition scored habituation/dishabituation task videos. Behavioral scoring was based on a predefined ethogram comprising two categories: (1) direct contact with the odor delivery vehicle and (2) attentive behavior, defined by forward movement, proximity to the vehicle, or odor. Time in direct contact and time attending to the vehicle or odor were summed within each epoch and expressed as a percent of total epoch duration (Equation 1: Time Spent Sniffing (%)). The percentage of time spent sniffing was then averaged across anosmic and sham conditions. Only animals confirmed to be behaviorally anosmic were included in the behavioral analysis. All testing was conducted at least 10 hours post-anesthetic induction and within 72 hours of whisker trimming and anosmia induction procedures.

$$TimeSpentSniffing(\mathrm{\%})=\left(\frac{\mathit{Time}\mathit{in}\mathit{Contact}+\mathit{Time}\mathit{Attending}}{\mathit{Total}\mathit{Epoch}\mathit{Duration}}\right)\times 100$$ Equation 1:

Video Analysis and Scoring Reaching Behavior:

Reaching behavior was recorded using three orthogonally positioned GoPro cameras (240 fps; GoPro, Inc, San Mateo, California). Videos were analyzed with BORIS behavioral scoring software (Friard and Gamba, 2016). Three trained independent observers, blind to the animals’ Sensory Condition, scored all trials. We adapted a four-point scoring system originally developed by Metz and Wishaw (2002). This scoring system includes four categories: Total Miss: 0 points are awarded when the reaching forepaw completely misses the cricket reward; Swipe: 1 point is awarded when the reaching forepaw contacts the cricket reward such that the cricket is knocked off the platform; Contact: 2 points are rewarded when the reaching forepaw contacts the cricket reward but is not successfully grasped and retrieved; Success: 3 points are awarded when the reaching forepaw grasps and retrieves the cricket (Equation 2: Weighted Error Formula ; Figure 2D). Although efforts were made to restrict animals to one reach per bout, many attempts sometimes occurred. We only scored the first attempt during each bout.

$$WeightedError=1-\frac{\mathit{Success}(3)+\mathit{Contact}(2)+\mathit{Swipe}(1)+\mathit{Total}\mathit{Miss}(0)}{3(\mathit{Total}\mathit{Number}\mathit{of}\mathit{Reaches})}$$ Equation 2:

Statistical Analysis

Inter-rater reliability Analysis

To assess inter-rater reliability among scorers, we computed intraclass correlation coefficients (ICC3, single fixed raters) across independent scorer’s blind to the experimental condition. Reliability was high (ICC3 = ICC3 = 0.96, p < 0.001). Linear regression analysis further demonstrated strong correspondence between scorers (R²adj = 0.868, F(1, 32) = 218.0, p < 0.001; Supplementary Figure 1A).

Analysis of Habituation/Dishabituation Task

To verify the effectiveness of the anosmia induction protocol, we analyzed behavior during a habituation/dishabituation task using a repeated-measures ANOVA statsmodels package in Python (Seabold and Perktold, 2010). Elapsed investigation time was compared between pre- and post-anosmia conditions (i.e., pre- and post-induction) within each subject. We used a repeated-measures ANOVA to test for differences within and across testing days and applied the Holm–Sidak method to correct for multiple comparisons. Animals ceased to respond to odor stimuli following anosmia induction (adjusted R² = 0.44, F (5) = 6.33, p = 0.012; pre-anosmia vs post-anosmia Odor 1: Tukey’s HSD: p < 0.0001; Supplementary Figure 2A). We observed a general decrease in interest in the vehicle stimuli when no odor was present after anosmia induction compared to when intact (Vehicle 1: Tukey’s HSD: p = 0.0267). Animals who underwent sham procedures continued to investigate the initial odor presentation at levels significantly greater than post-anosmia animals (Odor 1, Sham vs Post-Anosmia: Tukey’s HSD: p = 0.03; Supplementary Figure 2A), indicating that odor-directed investigatory behavior persisted following sham procedures. Reach-to-grasp performance also differed significantly between sham and anosmic conditions in both SC and EB animals (Paired t-test: t-statistic = −2.707, p = 0.024; Supplementary Figure 2B), indicating that olfactory deprivation impaired task performance beyond the effects of the infusion procedure alone.

Analysis of Reaching Data

We initially fit a linear mixed-effects model including all main effects and interactions among Visual Group (EB/SC), Sensory Condition (baseline/whisker trimmed/anosmic), and Lighting Condition (light/dark), along with Biological Sex and a random intercept for Animal ID (See Initial Model). Because each animal was tested repeatedly across sensory and lighting conditions, animal identity (ID) was included as a random intercept to account for within-subject correlations in weighted error. Model simplification was performed using backward hierarchical selection, and higher-order interactions were sequentially removed based on likelihood ratio and Akaike Information Criterion values. The final model retained interactions between Visual Group and Sensory Condition and between Visual Group and Lighting Condition, along with the main effects of these variables and Biological Sex (See Final Model)

Post-hoc pairwise comparisons of estimated marginal means (EMM) were conducted using the emmeans package (Lenth, 2025; version 1.11.1) in R (version 12.1), with Tukey’s correction applied to adjust for multiple comparisons. All contrasts reported are adjusted for biological sex to account for sexual dimorphism. The results of EMM contrast are found in Table 2.

Table 2:

Post hoc contrasts across light condition, sensory condition, and visual group.

Comparison Set	Contrast	Estimated Difference	SE	df	t	95% CI	p
Light Condition within Visual Group	SC: Light – Dark	−16.06	4.18	96.1	−3.842	[−24.40, −7.76]	0.0002***
	EB: Light – Dark	−3.22	4.52	107.6	−0.714	[−12.20, 5.73]	0.4769
Visual Group within Light Condition	Light: SC – EB	−16.76	4.85	47.5	−3.456	[−26.50, −7.01]	0.0012**
	Dark: SC – EB	−3.92	5.55	64.6	−0.706	[−15.00, 7.17]	0.4827
Sensory Condition within Visual Group	SC: Baseline – Anosmia	−18.03	5.50	114.4	−3.282	[−31.08, −4.98]	0.0039**
	SC: Baseline – Trimmed	−15.53	5.07	115.3	−3.064	[−27.56, −3.50]	0.0076**
	SC: Anosmia – Trimmed	2.51	5.68	98.7	0.441	[−11.02, 16.03]	0.8985
	EB: Baseline – Anosmia	−34.44	5.70	111.7	−6.046	[−47.97, −20.91]	< 0.0001***
	EB: Baseline – Trimmed	−19.59	5.00	104.4	−3.918	[−31.48, −7.70]	0.0005***
	EB: Anosmia – Trimmed	14.84	5.96	114.0	2.492	[0.70, 28.99]	0.0374*
Visual Group within Sensory Condition	Baseline: SC – EB	−3.52	5.23	67.1	−0.672	[−14.00, 6.93]	0.5039
	Anosmia: SC – EB	−19.92	6.99	89.9	−2.848	[−33.80, −6.02]	0.0055**
	Trimmed: SC – EB	−7.58	6.10	76.3	−1.244	[−19.70, 4.56]	0.2173
Sex Contrast	Female – Male	11.6	4.22	22	2.746	[2.84, 20.3]	0.0118*

Estimated marginal mean (EMM) contrasts were derived from a linear mixed-effects model. Values represent estimated differences in reaching error between conditions or groups.

Degrees of freedom were estimated using the Kenward–Roger method. Confidence intervals (CI) are reported at 95%. SC, sighted control; EB, early blind.

^*p < 0.05,

^**p < 0.01,

^***p < 0.001.

All remaining statistical analyses were conducted in Python using custom analysis scripts, including paired t-tests comparing reach-to-grasp errors between sham and anosmic conditions, repeated-measures ANOVA analyzing habituation/dishabituation behavior, Mann-Whitney U tests assessing differences in paw preference across visual conditions, and inter-rater reliability analysis evaluating scoring consistency between observers.

Initial Model:

$$\text{Weighted}\text{Error}~\text{Visual}\text{Group}\times \text{Sensory}\text{Condition}\times \text{Lighting}\text{Condition}+\text{Biological}\text{Sex}+(1|\text{Animal}\text{ID})$$

Final Model:

$$\text{Weighted}\text{Error}~\text{Visual}\text{Group}\times \text{Sensory}\text{Condition + Visual}\text{Group}\times \text{Light}\text{Condition}+\text{Biological}\text{Sex}+(1|\text{Animal}\text{ID})$$

Results

We trained early blind (EB) and sighted control (SC) adult opossums to perform a reach-to-grasp task (Figure 3). To determine the extent to which EB and SC opossums relied on nonvisual cues, we sequentially restricted whisker and olfaction-mediated sensory sampling by whisker trimming, and subsequently by inducing anosmia after the whiskers had regrown.

Figure 3: Images of reaching movements.

(A) Individual video frames of a sighted control and (B) an early blind short-tailed opossums during the reaching and grasping task. Images of sighted and blind animals were matched according to phase of the task, starting with the initial reach (top images), then the successful grasp, and lastly bringing the food to the mouth (bottom images).

Reaching and grasping in early blind and sighted opossums.

Short-tailed opossums appeared to be innately adept at reaching and grasping, as both EB and SC animals rapidly achieved low-weighted error rates during shaping. Although SC animals appeared to exhibit a delayed reduction in weighted error across shaping days compared to EB animals, learning trajectories did not differ between visual groups across the 5-day shaping period (Supplementary Figure 3A). No paw preference was observed in either group (Mann-Whitney U: 75.0, p = 0.169; Supplementary Figure 3B). A linear mixed model showed a significant effect of biological sex on reach-to-grasp performance (Male vs Female: p = 0.0073; See Table 2 for EMM contrast; Supplementary Figure 3C), with females committing more errors than males. These differences in weighted error by sex are a result of sexual dimorphism, where adult females, on average, are smaller than males (as indicated by lower body weight, smaller head size, and body length in females) (Bergallo and Cerqueira, 1994).

Effects of vision on reaching performance

To explore differences in reach-to-grasp performance between EB and SC opossums, reaching events were scored by two independent observers. We then fit a linear mixed effects model to predict how Visual Group (EB, SC), Light Condition (light, dark), and Sensory Condition (baseline, whisker trimmed, anosmia) affected reaching weighted error. Our model revealed a significant interaction between Visual Group and Light Condition (β = −12.83, SE = 5.90, t = −2.17, p = 0.032). Post-hoc estimated marginal means (EMM) analysis showed that SC opossums committed fewer errors than EB opossums when in the light (Table 2), but SC and EB animals performed similarly when in the dark (Table 2). Consistent with this pattern, SC animals showed worse performance in the dark (Table 2, Figure 4A). Unsurprisingly, lighting condition had no effect on performance in EB animals (Table 2; Figure 4A). These results suggest that visual cues substantially facilitate successful reach targeting for SC animals in this task, and that EB animals rely on other sensory modalities to compensate because vision is unavailable.

Figure 4: Reaching and grasping performance.

Box and whisker plots for percent weighted error for task performance in early blind animals (EB; red), and sighted controls (SC; blue). Estimated marginal means and standard errors are plotted as small boxes and whiskers adjacent to the main boxes. (A). SC opossums made significantly fewer errors in the light (white hatched) compared to the dark (black hatched), confirmed by estimated marginal means contrasts (EMM; small boxes beside whisker plots) (Light − Dark: p = 0.0002). SC opossums also made significantly fewer errors when in the light compared to EB opossums (SC – EB: p = 0.0012), while sighted and EB opossums performed similarly in dark conditions (SC – EB: p = 0.48). EB animals showed no significant change in performance across lighting conditions. (B) Plots show performance before (solid) and after whisker trimming (white dotted) and indicate that the lack of tactile input significantly increased error in both groups (linear mixed model: p = 0.0016, not illustrated in figure), with estimated marginal means confirming significant increases in error for both EB and SC animals (Baseline – Trimmed: p = 0.0005) and sighted animals (Baseline – Trimmed: p = 0.0076). Light and dark conditions were combined for this analysis. (C) Plots show performance before (solid) and after anosmia induction (black dotted). An increased error rate in both groups reflects the main effect of anosmia (linear mixed model: p = 0.0008, not illustrated in figure), with estimated marginal means confirming significant increased error in both sighted animals (Baseline – Anosmia: p = 0.0039) and EB animals (Baseline – Anosmia: p = 0.0001). A significant interaction between visual group (SC/EB) and pre-post anosmia was found (p = 0.031). EMM post-hoc analysis determined the direction of this effect: EB animals exhibited higher weighted error than sighted controls (SC/anosmic – EB/anosmic: p = 0.0055). Light and dark conditions were combined for this analysis. (D) Plots comparing performance between whisker trimmed (white dotted) and anosmic (black dotted) conditions. Early blind animals exhibited significantly greater targeting error under anosmia compared to whisker trimming (Anosmia – Trimmed: p = 0.0374). In contrast, sighted opossums showed no significant difference between the two conditions. In all plots boxes represent interquartile range, with inside bar representing the median whiskers showing 1.5 times the interquartile range.*p < 0.05, **p < 0.01, ***p < 0.001.

Reaching performance of early blind and sighted opossums is differentially affected by loss of input from other sensory systems

To examine the role of tactile input from the facial whiskers, and of olfaction in reach-to-grasp performance, we selectively restricted sensory input in these modalities in EB and SC opossums. Previous work from our laboratory has shown that EB opossums have alterations in receptive fields and response properties for neurons in the whisker representation of S1 and outperform SC animals in texture discrimination tasks (Ramamurthy et al, 2018, Ramamurthy et al, 2021; Figure 1C). Therefore, we trimmed the mystacial, genal, and submandibular vibrissae to determine their contributions to this reaching task in both SC and EB animals. As expected, we observed an overall increase in weighted error after whisker trimming in EB and SC animals (Supplementary Figure 3D). Our linear mixed effects model revealed a significant positive effect of whisker trimming on targeting error rates of EB and SC opossums (β = 15.53, SE = 4.80, t = 3.238, p = 0.0016; Figure 4B) with no evidence for a differential effect between EB and SC animals (Table 2). Thus, whisker trimming had a similar impact on performance for both SC and EB opossums under both light and dark conditions suggesting that information from the whiskers facilitates successful reach targeting in both groups.

To investigate the role of olfaction on reach-to-grasp, we induced anosmia and tested reach-to-grasp performance under light and dark conditions. Similar to whisker trimming, we observed an overall increase in average weighted error when comparing individual performance before and after anosmia induction (Supplementary Figure 3E). Our linear mixed effects model then revealed a significant increase in weighted error following anosmia induction across both visual groups (β = 18.03, SE = 5.20, t = 3.47, p = 0.0008; Figure 4C). Importantly, there was a significant interaction effect between Visual Group (EB /SC) and Sensory Condition (baseline/anosmic) (β = 16.40, SE = 7.50, t = 2.19, p = 0.031), indicating that the effect of anosmia on target error rate differed between EB and SC animals. Post-hoc estimated marginal means revealed that under anosmia, EB animals exhibited significantly larger target error rates than sighted controls (Table 2; Figure 4C). Thus, anosmia disproportionately decreased the performance of EB animals, suggesting that olfactory information substantially facilitates successful reach targeting for EB animals.

When comparing performance between anosmia and whisker-trimmed sensory manipulations, EB animals exhibited significantly greater targeting errors under anosmia compared to whisker trimming across lighting conditions (Table 2 ; Figure 4D), suggesting that EB opossums rely more on olfactory input than tactile input from the vibrissae. In contrast, although the performance of SC opossums was affected by loss of tactile input and olfactory input, there was no significant difference in their performance between the two manipulations (Table 2; Figure 4D), underscoring that their performance remains more stable across different sensory disruptions. These findings suggest that EB animals may rely more heavily on olfactory input to guide forelimb movements, whereas SC animals appear less affected by its loss, potentially relying more on visual cues.

Discussion

We examined the combined effect of early vision loss with tactile or olfactory deprivation in adults on reach-to-grasp performance. Early blind (EB) short-tailed opossums are capable of accurate reach-to-grasp movements, suggesting that the use of other senses can compensate for vision loss. Selective removal of tactile and olfactory input revealed that both EB and SC animals rely on these sensory modalities to retrieve prey, with olfaction and whisker touch contributing strongly to reach accuracy in EB animals.

Sensory-guided reaching and grasping in non-primate models

Primates are considered to be the reaching and grasping experts among mammals, especially those with opposable thumbs such as macaques, capuchin monkeys and humans (Christel and Fragaszy, 2000; Fox et al., 2019; Mitchell et al., 2024). Despite this emphasis on primate expertise, most reach-and-grasp studies use mice and rats (Whishaw et al., 2017; Naghizadeh et al., 2020; Parmiani et al., 2021; Quarta et al., 2022), and this study demonstrates that opossums are also innately skilled at reaching and grasping (Supplementary Figure 3A).

Although little is known about the natural predatory behavior of short-tailed opossums, laboratory studies suggest that they are at least, in part, dependent on prey caught with the forepaw, while rats often capture prey with their mouths (Ivanco et al., 1996). Direct comparisons of naïve reaching strategies in rats and opossums reveal that opossums are more precise, achieving higher success rates on first attempts, while rats appear less coordinated and require more training to achieve comparable success (Ivanco et al., 1996). Given that locomotor habit exerts strong selective pressure on limb morphology, these differences between opossums and rats may reflect the semi-arboreal niche that short-tailed opossums occupy in the wild compared to the terrestrial niche that brown rats would naturally occupy. Short-tailed opossums have specializations for arboreal locomotion that allow for greater forepaw dexterity (Ivanco et al., 1996), and there is evidence that single paw reaches may be selected for in a semi-arboreal niche. For example, comparative studies of arboreal (T. minor) and terrestrial (T. tana) tree shrews reveal large differences in their reaching and manual dexterity that correlates with the degree of habitual arboreal locomotion. Specifically, T. minor consistently relied on single-paw grasping strategies, while, T. tana was never observed grasping food rewards with the paws (Sargis, 2001). These comparisons indicate that adaptations to an arboreal lifestyle coincide with selection for single paw reaching and grasping capacities. These interspecies differences highlight the importance of species-specific motor and sensory adaptations when evaluating reach-to-grasp capabilities, making the short-tailed opossum a suitable single-paw model for understanding reaching behaviors and their underlying neural circuits.

The contribution of different senses to reaching and grasping.

All mammals actively guide their sensory effectors to control the sampling of information during exploration of their environment. It has been shown that movement repertoire can vary in complexity in a task-dependent manner, arguably to maximize the reward obtained over arbitrarily long timescales or task sequences. For example, whisking movement strategies are modulated during the exploration of a novel object (Gordon et al., 2014). This framework provides the foundation for examining how short-tailed opossums utilize whisker touch, olfaction, and vision during an ethologically relevant task such as reaching and grasping, and how the relative roles of these senses shift after early vision loss.

Here we show that in short-tailed opossums vision plays an important role for reaching since the error of SC opossums increased significantly in dark conditions compared to light conditions (Figure 4A). However, SC opossums could still perform the task in the dark, suggesting that although active visual sampling is beneficial, it is not necessary for success, and that the remaining sensory systems play a significant role in performance in SC animals and are the driving factors of performance in EB animals.

Previous studies in our laboratory found that EB opossums performed better than SC opossums in ladder-rung walking tasks, and compensate via tactile sensing with the mystacial whiskers, since trimming these whiskers decreased task performance (Englund et al., 2020). A subsequent study demonstrated that EB opossums had lower texture discrimination thresholds compared to SC opossums, an effect that disappeared after whisker trimming (Ramamurthy et al., 2021). In the current study, whisker trimming increased targeting error in both EB and SC animals, indicating that tactile input contributes to task performance and is one of the driving factors of behavioral compensation after the early loss of vision (Figure 4B). In EB humans, previous studies demonstrate that they have enhanced tactile acuity and vibrotactile perception (Wan et al., 2010; Radziun et al., 2023), and that the enhancement of tactilely mediated behavior may be facilitated by differences in the spatial tuning of tactile attention, which is more precise in EB versus sighted humans (Forster et al., 2007).

There are alterations in brain structure and function that may underlie the behavioral differences in tactile performance in EB and sighted individuals. In opossums, the increased reliance on whisker-mediated touch coincides with smaller receptive fields and sharper tuning of neurons in the whisker representation of S1 in EB animals (Ramamurthy and Krubitzer, 2018), the functional re-specification of “visual cortex” by the somatosensory and auditory system (Kahn and Krubitzer, 2002), and the emergence of novel cortical and subcortical inputs to V1 from somatosensory areas of the neocortex and nuclei of the thalamus (Karlen et al., 2006). In other animal models of early blindness such as anophthalmic mice, differences in functional, structural and anatomical organization of the brain are observed (Chabot et al., 2007, 2008; Massé et al., 2014) including a greater density of connections between S1 and “V1” (Charbonneau et al., 2012).

There are likely numerous anatomical and functional changes that account for compensatory behavior in EB humans. For example V1, extrastriate cortex, and portions of posterior parietal cortex are activated during Braille reading (Sadato et al., 1996; Buchel, 1998) suggesting that these areas are involved in tactile discrimination and active touch. The middle temporal complex, a region normally involved in processing visual motion and optic flow, is activated during tactile flow in EB individuals (Leo et al., 2012). This activation is proposed to play a central role in the perception of one’s movement relative to external objects and the environment (Ricciardi et al., 2007). Particularly relevant to the current study, functional imaging studies in EB individuals demonstrate that early visual areas (V1, V2 and V3) that would otherwise represent foveal vision instead encode reach direction, as do other cortical areas involved in visuospatial and visuomotor processing (Bola et al., 2023). Finally, indirect measures of cortical connectivity such as diffusion tensor imaging show that one such change is the presence of novel connections between V1 and S1 in blind individuals (Liu et al., 2007; Ptito et al., 2008; Klinge et al., 2010).

While there are no behavioral studies that describe the olfactory capabilities of short-tailed opossums, this study demonstrates that both EB and SC opossums rely on olfaction to target prey, and that this sense is critical in EB animals (Figure 4C, D). The extreme reliance on olfaction in this task is unsurprising since the sense of smell is the only way EB opossums can locate immobile prey that is out of whisker-range. In other animal models of early blindness (e.g. anophthalmic mice) studies show an increased performance in olfactory tasks such as retrieving buried food items compared to sighted mice, and these differences coincided with increases in the volume of olfactory structures such as pyriform cortex, the amygdala, and the glomerular and granular layers of the olfactory bulb (Touj et al., 2020, 2021). Studies in congenitally blind humans demonstrate lower odor detection thresholds compared to sighted individuals (Beaulieu-Lefebvre et al., 2011), increased sensitivity to odorants (Gagnon et al., 2014), and better performance on odorant localization tasks (Manescu et al., 2020). Functional magnetic resonance imaging studies show that these behavioral differences in EB humans are coincident with stronger activation of olfactory structures (e.g. orbitofrontal cortex, amygdala and hippocampus) as well an extensive activation of occipital cortex during task performance (Kupers et al., 2011).

Taken together, data in congenitally blind humans and other mammals indicate that there are remarkable differences in tactile and olfactory mediated behaviors including low-level processes such as sensory discrimination, to higher-level processes such as spatial localization and attention. Reaching and grasping objects require a representation of peri-personal space and the ability to distinguish self from the surrounding world (Ricciardi et al., 2007, 2017); and these higher-level capabilities require integrating vision with the remaining senses. The current study and previous studies demonstrate that in the absence of vision, the representation of peri-personal space is still present, and is likely generated by alterations in the size, functional organization, and connectivity of structures associated with the spared sensory modalities, as well as the co-opting of structures normally associated with visual processing.

Table 1:

List of Abbreviations

Abbreviation	Definition
EB	Early Blind
EMM	Estimated Marginal Means
P4	Postnatal Day 4
S1	Primary Somatosensory Area
SC	Sighted Control
V1	Primary Visual Area
V2	Second Visual Area
V3	Third Visual Area

Significance Statement.

Congenital sensory loss in humans alters the landscape used to navigate the world, making compensatory strategies mediated by the spared sensory systems essential for goal-directed behaviors such as reaching and grasping. Although these behaviors have been widely studied across species, the extent to which spared senses support their execution after congenital vision loss remains unclear. Here, we use the short-tailed opossum as a model of congenital blindness to quantify the contributions of whisker-mediated touch and olfaction to reaching performance. We show that both early blind and sighted opossums rely on whisker touch for reaching and grasping, but that olfaction plays a profound role in task performance in early blind opossums.