Abstract

A cohort of 13 adult macaques offered a unique opportunity to collect over several years manual dexterity data, from an initial learning phase in intact animals to a terminal phase of functional recovery after unilateral lesion of primary motor cortex (M1). Manual dexterity was assessed daily using the modified Brinkman Board task, yielding a total score given by the number of food pellets retrieved by one or the other hand from vertical and horizontal slots. A motor learning curve slope was established during the initial learning phase before reaching a stable performance with the dominant hand. Later, following contralateral M1 lesion, the manual dexterity score dropped to zero, before a progressive spontaneous functional recovery occurred, reaching a unique plateau of usually incomplete recovery. A recovery curve slope was calculated. In six of the 13 monkeys, a treatment aimed at enhancing the functional recovery of manual dexterity was applied, yielding a second plateau of recovery added to the first spontaneous recovery plateau. A recovery curve slope was also calculated for the second plateau. The hypothesis that steep initial motor learning is correlated with rapid and efficient functional recovery after M1 lesion was tested. In contradiction to this hypothesis, the data showed an inverse correlation with decreasing recovery curve slopes as a function of increasing learning curve slopes. This result suggests that the mechanisms underlying initial motor learning may be different from those mobilized for functional recovery after M1 lesion.

Introduction

As a result of practice or experience, motor learning of a specific, new motor task corresponds to a relatively permanent improvement in performance, until reaching a plateau of stable motor capacity. In a previous report (Kaeser et al., 2014), when confronted for the first time to a manual dexterity task, the modified Brinkman Board task, the motor learning properties of adult monkeys were illustrated and quantified. Young adult intact monkeys were trained to retrieve small food pellets from vertical and horizontal slots, using one or the other hand. As established for their dominant hand (Chatagny et al., 2013; Kaeser et al., 2014), the slope of the motor learning curve was obtained by dividing the gain of performance (number of pellets retrieved) by the time (number of days) until reaching the plateau of performance (Kaeser et al., 2014). Within an original population of 20 monkeys (Kaeser et al., 2014), there was a large variability of motor learning curve slopes, ranging from nearly zero (Mk-MO, no learning phase) to 0.42 (Mk-AT, abrupt learning phase). From this original group of 20 monkeys in which the learning phase for the modified Brinkman Board task was quantified, 13 of them were subjected later in their life to an experimental permanent unilateral lesion of the hand representation in the primary motor cortex (M1), as listed in Table 1. Immediately after the M1 lesion, the manual dexterity was totally suppressed (the score dropped to zero). In seven of those monkeys, in absence of any treatment (“untreated” monkeys), a spontaneous and progressive functional recovery from the M1 lesion took place (Figures 1A, B), until reaching a unique plateau of incomplete recovered performance, corresponding to a quite variable percent of functional recovery (Table 1). In the other six monkeys, the functional recovery was tentatively boosted via one or the other of two pilot therapies (Table 1). In these six “treated” monkeys, there was a first plateau of recovery corresponding to an initial spontaneous recovery, followed by a second plateau reflecting the recovery enhancement effect of the treatment (Figure 1C). The raw data for these 13 monkeys were published previously in detail, including illustrations of the manual dexterity performance with time, before and after the M1 lesion (Gindrat et al., 2025; Gindrat, 2015; Kaeser et al., 2011, 2014; Liu and Rouiller, 1999; Roux et al., 2025; Savidan et al., 2017; Wyss et al., 2013). As shown in Figure 1, a recovery curve slope can be calculated for both the first plateau and the second plateau of functional recovery.

Monkey IDLes. vol. (mm3)Treatment2 nd plateau% rec. T% rec. V% Rec. HScore T postScore V postScore H postInitial score TLearning slope (T)Recovery slope (T)Mk-BIa20.1NoneNo6897322116.54.529.50.03360.6571Mk-CEa112.8NoneNo386491091220.2740.0145MK-DGb32.2NoneNo6071431912620.50.1190.5Mk-DIc68.5NoneNo3979711.511120.50.4100.2097Mk-GEa48.7NoneNo4257111081180.05162Mk-ROa14.0NoneNo75878021131023.50.02880.4643Mk-ANd27.7NoneNo657055137690.07950.5217Mk-VA - 1st pl.20.0Control-likefirst plat50673213103240.09370.941Mk-MO - 1st pl.41.8Control-likefirst plat5689251917433.5-0.00371.6666Mk-JA - 1st pl.20.5Control-likefirst plat63635417107230.01990.9Mk-JO - 1st pl.30.0control-likefirst plat3456011.510024.50.1210.6316Mk-CA - 1st pl.22.0Control-likefirst plat50762515133.5120.05980.2429Mk-LO - 1st pl.19.1Control-likefirst plat656467179812.50.08270.5313Mk-VAe20.0Anti-Nogo-AYes87877322.5138240.09370.1886Mk-MOe41.8Anti-Nogo-AYes7684562616933.5-0.00370.57Mk-JAa20.5ANCEYes989410026.51513230.01990.3333Mk-JOa30.0ANCEYes5989252016424.50.1210.2962Mk-CAd22.0ANCEYes67716120128.5120.05980.0943Mk-LOd19.1ANCEYes92939224131112.50.08270.0787

List of monkeys with their ID (leftmost column) and the volume of the primary motor cortex lesion (2nd column).

The 3rd column from the left indicates the corresponding therapy or treatment: none is for “untreated” (no treatment) monkeys (n = 7), “control-like” is for the first plateau (spontaneous recovery) in the “treated” monkeys (see Figure 1C; n = 6); overall, the number of “control” data points is 13 (7+6). Two monkeys (Mk-VA and Mk-MO) were treated with an anti-Nogo-A antibody, whereas four monkeys were subjected to the ANCE cellular therapy: overall, the pooled treated monkeys yielded six “treated” data points corresponding to their second plateau of functional recovery. Note that the learning curve slope (second column from right), before M1 lesion, obviously was the same for the six treated monkeys for their first or second plateau data line. The 4th column from the left indicates the absence (No) or presence (Yes) of a second plateau of functional recovery of manual dexterity; in case of the presence of a second plateau of recovery, the first plateau was considered as spontaneous recovery [see also (Rouiller, 2026)], thus yielding an additional “control-like” (i.e., no treatment) data point for the corresponding monkey (lines with gray characters). Data points in black or gray thus represent spontaneous functional recovery, while data points in red or green represent treatment enhanced functional recovery (anti-Nogo-A antibody or ANCE, respectively). The nine rightmost columns list the nine behavioral parameters considered in the analysis (see text). In these nine columns headings, “T” is for the Total number of slots, “V” is for the number of Vertical slots, and “H” is for number of Horizontal slots. “% Rec.” is for the percentage of functional recovery. The scores post-lesion (“post”) are given by their median values, either at first plateau (n = 13) or at second plateau (n = 6), separately for the total number of slots (“T”), the vertical slots (“V”), and the horizontal slots (“H”). The “Initial Score T” computed for the total number of slots corresponds to the score estimated for each monkey before starting the motor learning phase in the intact state [derived and rounded based on Kaeser et al. (2014): Table 1]. The learning curve and the recovery curve slopes were computed for the total number of slots (“T”). Original data (pre-lesion plateau and post-lesion plateau) derived from:

cGindrat, 2015; Gindrat et al., 2025,

eWyss et al., 2013. The learning curve slope data are those originally reported in Kaeser et al. (2014) (see Table 1). The recovery curve slopes were calculated as illustrated in Figure 1 and listed here based on the behavioral data (total number of slots) previously published (a–e above). The recovery curve slopes are the only new data introduced in the present report, but they were computed based on previously published data (a–e above). In our large collection of monkeys, two more monkeys (Mk-LA and Mk-SL) were subjected to M1 lesion and anti-Nogo-A antibody treatment. However, they were not included in the present report because the initial motor learning data were not available. Sex: Mk-DI, Mk-GE, Mk-AN, Mk-CA, Mk-LO were female macaques whereas Mk-BI, Mk-CE, Mk-DG, Mk-RO, Mk-VA, Mk-MO, Mk-JA, Mk-JO were males. Age: Most monkeys’ age at M1 lesion time ranged from 3.5 to 5 years old, except the older monkeys Mk-DI (9.5 y), Mk-DG (9.5 y), Mk-AN (14 y), Mk-CA (11 y) and MK-LO (11.5).

Two hypotheses to be tested. (A) Cartoon illustrating the typical time course of manual dexterity score for an adult “untreated” macaque monkey (“monkey a”), based on the modified Brinkman Board task executed with the dominant hand, including the initial motor learning phase, with a steep learning curve slope (“Ln. Sl. a”). The end of the learning phase is characterized by a plateau. Then, a unilateral experimental M1 lesion contralateral to the tested hand took place (purple vertical line), which provoked a total loss of manual dexterity during a few weeks. It was followed by a spontaneous, progressive functional recovery, until reaching a single plateau of recovery. The dashed line represents the corresponding recovery curve slope (“Rc. Sl. a”). In absence of treatment, the plateau remained stable for months to years. The vertical arrow represents the recovered post-lesion score. (B) Same as in panel (A), but for a gentle learning curve slope represented by an untreated “monkey b.” Note that both slopes (learning curve slope and recovery curve slope) are clearly less steep than for “monkey a.” Seven monkeys included in the present study corresponded to such a time course, as depicted in panels (A,B), with various learning and recovery curve slopes (Table 1). The hypothesis 1 tested in the present report argues that the initial motor learning slope is correlated with the slope of functional recovery from a M1 lesion, possibly following the relationship between learning curve slope and recovery curve slope as depicted in panel (D). As a consequence, the hypothesis 2 argues that in case of “steep initial motor learning” a higher recovered score of manual dexterity after M1 lesion will be reached than in case of “gentle initial motor learning.” (C) Same as in panels (A,B), but for a M1 lesioned monkey subjected to a treatment (either anti-Nogo-A antibody or ANCE autologous cellular therapy; n = 6 monkeys). The M1 lesion is indicated by the vertical purple line, followed by the treatment onset (red arrow pointing down). Following the total loss of manual dexterity, a first plateau of functional recovery took place, reflecting an initial spontaneous recovery. This first plateau of recovery was used to derive six additional data points to the seven data points illustrated in panels (A,B). Overall, 13 data points were representative of the spontaneous recovery post-M1 lesion [Table 1; see also (Rouiller, 2026) for more detail]. As a result of treatment, these six monkeys exhibited a second plateau of functional recovery, representing the effect of the treatment (red curve). As for the first plateau, a recovery curve slope (“Rc. Sl.”) and a recovery score (red arrow pointing up) can be derived specifically for the second plateau (Table 1), corresponding overall to six “treatment” data points. (D) Expected correlation between the recovery curve slope and the initial learning curve slope, for both plateau 1 and plateau 2 (see above).

The initial motor learning curve slopes and the subsequent functional recovery curve slopes from M1 lesion are listed for the 13 monkeys in Table 1. Capitalizing on such a rather unique opportunity to confront, for the very same motor task (modified Brinkman Board task) and in the same individual, the motor learning curve slope in the intact state and the functional recovery curve slope post-M1 lesion, one can address the following question: is there a relationship between initial learning slope and functional recovery slope post-M1 lesion?

One may be tempted to expect a correlation between the two slopes, with probably “steep initial motor learning” in the intact state associated to a more rapid functional recovery in case of M1 lesion occurring later in life, as illustrated in Figure 1D (hypothesis 1). Furthermore, “steep initial motor learning” may be correlated with a higher plateau of functional recovery after M1 lesion late in life (hypothesis 2 in Figure 1).

Materials and methods

Manual dexterity was investigated and quantified in adult monkeys (macaca fascicularis) on the basis of our modified Brinkman Board task, as previously reported in several articles from this laboratory (Chatagny et al., 2013; Gindrat et al., 2025; Gindrat, 2015; Hoogewoud et al., 2013; Kaeser et al., 2011, 2014; Liu and Rouiller, 1999; Rouiller et al., 1998; Rouiller, 2026; Roux et al., 2025; Savidan et al., 2017; Schmidlin et al., 2011; Wyss et al., 2013) and derived from previous versions of the task (Brinkman and Kuypers, 1972, 1973; Brinkman, 1984). Briefly, a manual dexterity daily score was calculated, given by the number of pellets successfully retrieved from vertically oriented slots and from horizontally oriented slots during the first 30 s of the test, yielding a “vertical” score and a “horizontal” score; furthermore, a “total” score was derived, given by the sum of the vertical and horizontal scores (Supplementary Video 1). After M1 lesion, following a transient total loss of manual dexterity (as illustrated in Figure 1), a progressive functional recovery was observed, reaching a single plateau in “untreated” monkeys, while “treated” monkeys exhibited two plateaus of recovery. For each plateau, a percentage of functional recovery was calculated, dividing the post-lesion median score by the pre-lesion median score × 100. The variability of the daily behavioral scores established with the modified Brinkman Board task was quantitatively reported in a recent report (Rouiller, 2026). The precise housing conditions of the monkeys in the animal facility were reported in detail earlier (Kaeser et al., 2014); see also www.unifr.ch/spccr/about/housing.

At a relatively young age [range 3–7 years old; weight 2.5–5.2 Kg; see (Kaeser et al., 2014)], the monkeys were first exposed to the modified Brinkman Board task, exhibiting for most of them a motor learning phase, characterized by a progressive increase in performance, until reaching a plateau of stable performance. This learning phase was used to derive for each monkey a learning curve slope (number of pellets increase divided by the duration in days of the learning phase), as illustrated earlier (Kaeser et al., 2014) and reminded here in Table 1 for the 13 monkeys included in the present study. Based on their behavioral data previously published (Gindrat et al., 2025; Gindrat, 2015; Kaeser et al., 2011; Liu and Rouiller, 1999; Roux et al., 2025; Savidan et al., 2017; Wyss et al., 2013), for the present study an original recovery curve slope was calculated as illustrated in Figure 1, for both the unique/first plateau and the second plateau of functional recovery (listed in Table 1 for each monkey).

In the 13 monkeys subjected to a unilateral M1 lesion, seven exhibited a spontaneous functional recovery, as they were not subjected to any treatment (“untreated” monkeys; see Table 1). In contrast, six monkeys were “treated,” either with an anti-Nogo-A antibody (Freund et al., 2006, 2007, 2009; Hamadjida et al., 2012; Hoogewoud et al., 2013; Schwab, 2004, 2010; Wyss et al., 2013) or with the ANCE autologous cellular therapy (Bloch et al., 2014; Borgognon et al., 2017, 2019; Brunet et al., 2005, 2009; Kaeser et al., 2011; Roux et al., 2025). The M1 permanent lesions, induced chemically by infusion of ibotenic acid, were histologically reconstructed, represented on lateral views of the brain (Gindrat, 2015; Kaeser et al., 2011; Liu and Rouiller, 1999; Roux et al., 2025; Savidan et al., 2017; Wyss et al., 2013) and, finally, a lesion volume was calculated for each monkey (Table 1). All these methods (M1 lesion procedure, lesion reconstruction and treatments) were described in great detail earlier (see articles from this laboratory mentioned above) and are therefore not repeated here in the current brief research report.

Results

Table 1 lists the data relevant for the hypotheses to be tested in the present report (Figure 1). Most of these data (columns 2–12 from the left in Table 1) were published previously (Chatagny et al., 2013; Gindrat et al., 2025; Gindrat, 2015; Hoogewoud et al., 2013; Kaeser et al., 2011, 2014; Liu and Rouiller, 1999; Rouiller et al., 1998; Rouiller, 2026; Roux et al., 2025; Savidan et al., 2017; Schmidlin et al., 2011; Wyss et al., 2013). The new data provided here consist of the “recovery curve slopes” (rightmost column in Table 1). In order to test the hypothesis 1 (see Figure 1D), the recovery curve slope was plotted as a function of the corresponding motor learning curve slope (Figure 2A), separately for the first plateau data (blue dots) and the second plateau data (brown dots). In contrast to the hypothesis 1, the data show for both plateaus that, in the same monkey, steep recovery curve slopes were rather correlated with gentle learning slopes, and then recovery curve slopes tended to decrease as a function of increasing learning curve slopes (Figure 2A). This overall tendency was best fitted with a logarithmic function, exhibiting a statistically significant coefficient of inverse correlation (p < 0.05 for the first plateau data and p < 0.01 for the second plateau data; see Supplementary Table 1 for precise p-values).

Modified Brinkman Board (BB) data (part 1). (A) The recovery curve slope was plotted as a function of the corresponding initial learning curve slope in the same monkey, separately for the 13 “unique/first plateau” data points (blue dots) and the six “second plateau” data points (brown dots). The manual dexterity scores used to derive the learning curve slope and the recovery curve slope were those established for the total number of slots retrieved in 30 s (see section “2 Materials and methods”). The data points were fitted with a logarithmic function (dashed blue and brown curves), with the corresponding coefficient of correlations (r = ) and the statistical significance or not of the p-values (n.s. is for p > 0.05). See Supplementary Table 1 for the precise p-values. (B) The recovered post-lesion score (total number of slots) was plotted as a function of the initial learning curve slope, fitted with a regression line. Otherwise, same conventions as in panel (A). (C) The percentage of functional recovery (total number of slots) was plotted as a function of the initial learning curve slope, fitted with a regression line. Otherwise, same conventions as in panel (A). (D) The recovery curve slope was plotted as a function of the M1 lesion volume, fitted with a regression line. Otherwise, same conventions as in panel (A).

To test the hypothesis 2 (Figure 1), the post-lesion motor performances were confronted to the motor learning curve slopes, considering the median total score post-lesion (Figure 2B) or the percentage of functional recovery for the total scores (Figure 2C). In contrast to the hypothesis 2, the total scores post-lesion tended also to be inversely correlated to the motor learning curve slopes (Figure 2B); this was true for both the first plateau data (blue dots) and the second plateau data (brown dots). The two subpopulations data were fitted with a regression line, exhibiting a statistically significant coefficient of correlation (p < 0.05) for the first plateau data, but not for the second plateau data (Figure 2B and Supplementary Table 1). In contrast to clinical studies in which only the post-lesion behavioral data are available, the advantage of the present monkey model is to have access to the pre-lesion behavioral data. As a result, the functional recovery of manual dexterity can be expressed by a percentage value, by dividing the post-lesion score by the pre-lesion score. As shown in Figure 2C, the percentage of functional recovery for the total score tended also to be inversely correlated to the motor learning curve slopes (coefficient of correlation statistically significant only for the first plateau data: see Supplementary Table 1). Nevertheless, the Figures 2B, C indicate that both parameters reflecting the post-lesion performance yielded consistent data, contradicting the hypothesis 2. Although comparable tendencies to the total scores (Figures 2B, C) were observed when plotting as a function of learning curve slopes the post-lesion score or percentage of functional recovery for the horizontal scores or the vertical scores separately (Table 1), the corresponding coefficients of correlation were not statistically significant (not shown).

Finally, one may ask whether the volume of the M1 lesion impacts on the recovery curve slope, with a tentative expectation that steep recovery curve slope and best recovered motor performance should be present when the M1 lesion was modest in size. The data are presented in Figure 2D, where the recovery curve slopes were plotted as a function of the M1 lesion volumes, again separately for the first plateau data and the second plateau data. For the first plateau data (blue dots), the above prediction was not verified, with a poor correlation between these two parameters. The second plateau data, in contrast, exhibited a tendency rather opposite the above prediction, though not statistically significant (p > 0.05; see Supplementary Table 1).

The data presented above focused on the here newly introduced parameter of recovery curve slope after M1 lesion and its relation with the initial motor learning curve slope when the monkey was intact and exposed for the first time to the modified Brinkman Board task (Figure 2A). The motor learning curve slope was quite variable among the monkeys (Table 1), indicating significant interindividual differences with respect to margin of progression during the motor learning process. However, in the context of a supplementary analysis, one may also consider the early intrinsic motor capacity of each monkey before the initial motor learning phase, which was given by an “early in