Division of Labor: A Democratic Approach to Understanding Manual Asymmetries in Non-Human Primates

A consequence of the ‘gold rush’ like hunch for human-like handedness in non-human primates has been that researchers have been continually analyzing observations at the level of the population, ignoring the analysis at the level of an individual and, consequently, have potentially missed revelations on the forms and functions of manual asymmetries. Recently, consecutive studies on manual asymmetries in bonnet macaques, Macaca radiata [Mangalam et al., 2014a; Mangalam et al., 2014b] revealed both the functional and the adaptive significance of manual asymmetries respectively, and pointed towards the division of labor as being the general principle underlying the observed hand-usage patterns. We review the studies on manual asymmetries in capuchin monkeys, Cebus spp. and argue that the observed hand-usage patterns might reflect specialization of the two hands for accomplishing tasks that require different dexterity types (i.e., maneuvering in three dimensional space or physical strength). To this end, we do a step-by-step analysis of the various tasks used in the studies on manual asymmetries in capuchin monkeys, wherein we: (a) analyze the different manual tasks that have been used to study manual asymmetries in non-human primates on the basis of the attributes such as the number of hands required to solve a given task (i.e., unimanual, pseudo unimanual, or bimanual) and the spatiotemporal progression of manual actions (i.e., sequential or concurrent). (b) Determine the forms and functions of manual asymmetries that these tasks can potentially elicit within the broader scope of the behavioral repertoire of an individual, a population, or a species. (c) Qualify the scope of the inter-individual, -population, or -species comparisons. We then describe the division of labor as a general principle underlying manual asymmetries in non-human primates, and propose experimental designs that would elaborate the forms and functions of manual asymmetries in non-human primates, and the associated adaptive value.

being the general principle underlying the observed hand-usage patterns. We 23 review the studies on manual asymmetries in capuchin monkeys, Cebus spp. and 24 argue that the observed hand-usage patterns might reflect specialization of the two 25 hands for accomplishing tasks that require different dexterity types (i.e., 26 maneuvering in three dimensional space or physical strength). To this end, we do a 27 step-by-step analysis of the various tasks used in the studies on manual 28 asymmetries in capuchin monkeys, wherein we: (a) analyze the different manual 29 tasks that have been used to study manual asymmetries in non-human primates on 30 the basis of the attributes such as the number of hands required to solve a given 31 task (i.e., unimanual, pseudo unimanual, or bimanual) and the spatiotemporal 32 progression of manual actions (i.e., sequential or concurrent). (b) Determine the 33 forms and functions of manual asymmetries that these tasks can potentially elicit 34 within the broader scope of the behavioral repertoire of an individual, a population, 35 or a species. (c) Qualify the scope of the inter-individual, -population, or -species 36 comparisons. We then describe the division of labor as a general principle 37 underlying manual asymmetries in non-human primates, and propose experimental 38 designs that would elaborate the forms and functions of manual asymmetries in 39 Introduction 44 Approximately 90% humans preferentially use the right hand to perform complex 45 manual actions [Raymond and Pontier, 2004]. In order to understand the adaptive 46 value of this population-level right-handedness, which is peculiar to humans, it is 47 important to understand the evolutionary origin of manual asymmetries, in humans 48 as well as in their phylogenetic relatives, the non-human primates. Manual 49 asymmetries of some kind or the other are almost ubiquitous among the non-50 human primates. However, for a long time the population-level lateral bias in hand 51 usage in non-human primates remained equivocal; considering that the exogenous 52 factors, such as the initial position of a stimulus with respect to a subject, body 53 posture of the subject, etc. might influence hand usage, researchers considered 54 manual asymmetries in non-human primates to be analogous and not homologous 55 to manual asymmetries in humans. Regardless of such an ambiguity, hand 56 preference in non-human primates has been hypothesized to have evolved owing to 57 functional and morphological adaptations to feeding in arboreal contexts [Bradshaw 58 and Rogers, 1993; Papademetriou et al., 2005;Ward and Hopkins, 1993]. 59 60 As opposed to the prevailing ideas on population-level right-hand preference in 61 humans, MacNeilage et al. [1987] argued that human-like population-level lateral 62 bias in hand usage is evident in non-human primates, and proposed the postural 63 origins theory. According to the postural origins theory, among non-human 64 primates initially the left hand became specialized for visually guided movements, 65 and the right hand became specialized for postural support. Subsequently, in non-66 human primate species that adopted a relatively more terrestrial lifestyle, the right hand became more specialized for physical manipulation than for postural support, 68 owing to the decreasing demands on the right hand to support vertical posture. 69 However, the postural origins theory fails to describe why initially the left-hand 70 (and not the right hand) became specialized for visually guided reaching, and more 71 importantly, how a population-level right-handedness evolved during the transition 72 from monkeys to apes to humans [McGrew and Marchant, 1997]. Overall, the 73 postural origins theory incorporates the physical constraints on hand usage imposed 74 by the body posture, but does not explain the variations in hand-usage patterns, 75 corresponding to the novelty and the spatiotemporal scale of the manual actions. 76

77
In the earlier studies on manual asymmetries in non-human primates, terms such 78 as 'task complexity' and 'task demands' were used without ever being 79 comprehensively defined. For example, complexity of a reaching-for-food task was 80 measured in terms of the number of steps preceding the terminal act of reaching 81 for food, with almost no reference to the precision of movement in any of the 82 manual actions. This made it difficult to draw any conclusions with regard to the 83 forms and functions of manual asymmetries in non-human primates. Subsequently, 84 based on the perspective put forward by MacNeilage et al. [1987], while 85 simultaneously acknowledging the possibility that hand-usage patterns might vary 86 with novelty and the spatiotemporal scale of the manual actions, as indicated by 87 the previous studies on hand-usage patterns in non-human primates, Fagot and 88 Vauclair [1991] put forward the task complexity theory. The task complexity theory 89 proposes: (a) low-level tasks (i.e., tasks involving cognitively less demanding 90 actions that are practiced frequently) elicit symmetrical hand-usage patterns at the 91 level of the population and manual preferences at the level of an individual, not 92 necessarily indicative of any kind of specialization. (b) High-level tasks (i.e., tasks 93 involving cognitively more demanding manual actions that are practiced rarely) 94 elicit asymmetrical hand-usage patterns at the level of the population, likely to be 95 indicative of some kind of cognitive specialization. They also argued that 96 inconsistencies in directional biases arise owing to the diversity in the tasks used to 97 elicit manual asymmetries and the cognitive processes involved in solving them. 98 Overall, these two types of tasks, low-level and high-level, elicit two different types 99 of lateralization, hand preference and manual specialization. 100

101
Since the conception of the postural origins theory and the task complexity theory, 102 there have been a plethora of studies on manual asymmetries in non-human 103 primates with titles like "Laterality of hand functions in…," "Hand preferences in 104 different tasks in…," "Consistency of hand preference across low-level and high-level 105 tasks in…," "Hand preferences in unimanual and coordinated-bimanual tasks by…," 106 "Posture and reaching in…," etc. These studies generally have not independently 107 considered the constraints consider by the task complexity theory and the postural 108 origins theory. The task complexity theory incorporates the physical constraints 109 imposed by tasks, whereas the postural origins theory incorporates the physical 110 constraints imposed by body postures. These different types of physical constraints, 111 however, may not necessarily elicit mutually consistent hand preferences. They 112 have focused essentially on hand preference (i.e., the relative incidence of the use 113 of either hand for responding) as the primary measure to assess manual 114 asymmetries, with almost no reference to the forms and functions. Moreover, they 115 have continually ignored several individual-specific traits, such as the feeding 116 ecology and niche structure, and task-specific characteristics, such as the 117 spatiotemporal requirements of the task, which might together influence hand-118 usage patterns. In such a situation, conclusions drawn from studies incorporating 119 variable methodologies and task requirements, and not incorporating the 120 differences between individuals, populations, or species, are likely to be misleading. Manual asymmetries did not first evolve in primates, but hemispheric specialization 184 preceded manual symmetries instead, or in other words, evolved as a by-product of 185 a more fundamental cerebral asymmetry affecting sensorimotor functioning 186 [Witelson, 1988]. Accordingly, tasks that are likely to challenge the differential 187 abilities of the two hemispheres are more likely to elicit manual asymmetries: hand 188 preference, that is, the preferential usage of one hand to perform a unimanual task 189 or to execute the most complex action while performing a bimanual task, or hand 190 performance, that is, differential performance of the two hands in solving the same 191 task [Fagot and Vauclair, 1991]. In the manual preference paradigm, repetitive 192 presentations of a given task produce individual scores of right-and left-hand uses. tasks, we use the word 'pseudo' before unimanual). 246 247 Bipedal (pseudo) unimanual reaching-for-food tasks can only be solved using both 248 hands and in no less than two or three steps: (P1) two-step process: step 1: 249 setting one hand, hand-1 (i.e., either left or right hand), free from maintaining 250 quadrupedal posture and using it to hold a high-rise structure (this action is 251 physically demanding as the body is lifted/pulled upwards) while maintaining 252 tripedal posture using the other hand, hand-2; step 2: setting the other hand, 253 hand-2, free from tripedal posture and using it to reach for food while maintaining 254 bipedal posture using the other hand, hand-1. (P2) Three-step process: step 1: 255 setting one hand, hand-1, free from maintaining quadrupedal posture and using it to hold a high-rise structure (as mentioned above, this action is physically 257 demanding as the body is lifted/pulled upwards) while maintaining tripedal posture 258 using the other hand, hand-2; step 2: setting the other hand, hand-1, free from 259 tripedal posture and using it to hold the high-rise structure; step 3: using one hand, 260 (P1a) hand-1 (in which case the sequence is functionally similar to the previous 261 one) or (P2b) hand-2, to reach for food. 262 263 These sequences of manual actions involve both hands, following the principle of 264 division of labor, that is, one hand is used to perform the actions demanding 265 relatively more physical strength (e.g., lifting/pulling the body) and the other hand 266 is used to perform the actions demanding more sophistication (e.g., making 267 precision grips or maneuvering in three-dimensional space). However, studies on 268 hand preference in capuchins have almost never reported the stepwise usage of 269 the two hands for solving bipedal (pseudo) unimanual reaching-for-for-food tasks 270 as described above, restricting their data collection and analysis only to manual 271 actions that are directly associated with prehension. Comparative assessment of 272 hand preference in the quadrupedal and bipedal (pseudo) unimanual reaching-for-273 food tasks, as reported by Spinozzi et al. [1998] and Westergaard et al. [1997Westergaard et al. [ , 274 1998], demonstrates that capuchins consistently use one hand for prehension in 275 both types of tasks, which is possible only while following either the two-step 276 process (i.e., P1) or the second of the three-step processes (i.e., P2b) for solving 277 bipedal (pseudo) unimanual reaching-for-food tasks. Haptic discrimination has been found to be more difficult that visual discrimination 347 in non-human primates (see, for example, Wilson [1965]  Westergaard and Suomi, 1994b] (another version may involve using a sponge 369 [Westergaard and Suomi, 1993a]) or a bipedal posture [Lilak and Phillips, 2008;370 Westergaard, 1991;Westergaard et al. , 1998a]; another tool-using task is nut-371 cracking that involves coordinated bimanual handling of stones to crack nuts 372 [Westergaard and Suomi, 1993b;Westergaard and Suomi, 1996]. It is important to 373 note here that the above probing/tool-using tasks are similar in terms of the 374 number of hands required to solve the task (i.e., unimanual, pseudo unimanual, or 375 bimanual) and the spatiotemporal progression of manual actions (i.e., sequential or 376 concurrent) except for the fact that they involve an extension of the body, 377 controlling which requires finer finger adjustments through response-produced 378 feedback. Thus, functionally similar to simple reaching-for-food tasks, probing/tool-379 using tasks are likely to prove helpful only if the form of manual asymmetries (i.e., 380 with respect to grip type) is considered. 381

(vii) Spontaneous Tasks 383
Hand-usage patterns in tasks such as grooming [Fragaszy and Mitchell, 1990], 384 maternal cradling and infant positioning [Hopkins, 2004;Panger and Wolfe, 2000;385 Westergaard et al., 1999]are more likely to be influenced by the specialization of 386 the two hands for more common activities such as feeding than these tasks 387 themselves. For example, a female capuchin which has its left hand specialized for 388 fine finer adjustments or maneuvering in three dimensional space and its right hand 389 specialized for physical support is more likely to use its right hand for maternal 390 cradling and infant positioning just to keep its left hand free for the usual feeding 391 activities (as they require more sophisticated manual actions). However, studies merely describe the hand used for these activities without considering the forms 393 and functions of the associated manual asymmetries. 394 395

Forms and Functions of Manual Asymmetries 396
The corticomotoneuronal connections innervating the hands regulate the timing and 397 precision of the muscular forces required for fine finger adjustments through 398 response-produced feedback (see, for example, Porter [1985]). It follows from this 399 fact that actions with finer sequential finger movements are more likely to elicit 400 manual asymmetries than simpler actions, as Elliott and Chua [1996]  Sapajus xanthosternos: the red howlers, which habitually use the mouth to obtain 426 food, selectively took part in the reaching-for-food tasks and also exhibited stronger 427 hand preferences than the yellow-breasted capuchins in the tasks that were 428 relatively simple to solve. However, differences in the strength of hand preference 429 diminished with the increasing complexity of the reaching-for-food tasks, that is, 430 the relatively more complex tasks were perceived as equally complex by both the 431 red howlers and the yellow-breasted capuchins. Both these observations 432 demonstrate that the feeding ecology and niche structure influence hand-usage 433 patterns, bringing about the differences in hand preference out of the contingent 434 nature of the complexity of a task. Thus, manual asymmetries in non-human 435 primates should be investigated not just in isolation, but within the broader scope 436 of the behavioral repertoire of an individual, a population, or a species. opposability [Napier and Napier, 1967]. For a long time it was thus held, that no 485 New World monkey species could grasp objects with precision [Bishop, 1964;486 Napier, 1993;Napier and Napier, 1967]. However, comparative behavioral studies demonstrated that capuchins stand out from other platyrrhine species because of 488 their (a) high degree of manual dexterity [Fragaszy, 1986;Lacreuse and Fragaszy, 489 1996;Panger, 1988], (b) frequent use of precision grips that mainly involve lateral 490 aspects of digits for picking up small objects [Christel and Fragaszy, 2000;Costello 491 and Fragaszy, 1988;Spinozzi et al. , 2004], and (c) capacity to perform relatively 492 independent movements of the digits [Christel and Fragaszy, 2000;Costello and 493 Fragaszy, 1988]. 494 495 Anatomical and physiological features of the neural substrate that control manual 496 actions might explain the high manual dexterity in capuchins. Capuchins can act out 497 highly fractionated movements of the fingers/digits owing to the large number and 498 extension of the corticomotoneuronal connections that innervate the hand 499 [Kuypers, 1981;Lemon, 1993;Muir and Lemon, 1983;Shinoda et al., 1981], as 500 observed in humans and chimpanzees [Bortoff and Strick, 1993]. Moreover, studies 501 reported that the individuals that preferentially used the right hand to reach for 502 food in a concurrent bimanual tube task, exhibited a greater leftward bias of the 503 anterior cerebellum [Phillips and Hopkins, 2007], and had a shallower central sulcus 504 [Phillips and Sherwood, 2005] as well as a smaller overall corpus callosum in the 505 contralateral hemisphere , compared to those that 506 preferentially used the left hand or did not show hand preference; although there 507 was no difference in the size of the left-frontal petalia between the two [Phillips and 508 Sherwood, 2007]. 509 510 A few studies investigated manual asymmetries with respect to the control and 511 movement of the fingers/digits in capuchins. Christel and Fragaszy [2000] reported 512 that the individuals did not exhibit considerable patterns in hand preference or hand 513 performance with respect to the power or precision grips used to grasp currants 514 and grapes lying on a tray. Spinozzi et al. [2004] reported that the individuals 515 preferentially used one hand to grasp a food item fixed on a tray, and did not show 516 any difference in performance with respect to the power or precision grips, but 517 extracted the food faster with the preferred hand than the non-preferred hand with 518 respect to the precision grips (and not with respect to the power grips). Spinozzi et 519 al. [2007] reported that the individuals preferentially used one hand to retrieve a 520 raisin from a transparent hollow tube fixed horizontally to the upper end of a 521 vertical metal bar, and extracted the food faster with the preferred hand than the 522 other hand. Whereas these findings indicate that precise control/movement of the 523 fingers/digits are more likely to elicit manual asymmetries than the imprecise ones, 524 there are problems with the experimental setups. 525 526 If, suppose, one hand is specialized for manual operations that primarily involve 527 physical strength and, therefore, require power grips, and the other hand is 528 specialized for those that involve maneuvering in three-dimensional space and, 529 therefore, require precision grips, a manual operation that primarily requires either 530 one or the other of the two forms and functions of the hand is likely to influence 531 hand-usage patterns with respect to a particular type of grip as well as grip-532 formation patterns with respect to a particular hand. The three studies- Christel  between the two hands with respect to the power grips, presenting a distorted and 539 partial picture of manual asymmetries. 540

541
We propose an experimental design to unambiguously determining the forms and 542 functions of manual asymmetries in non-human primates. One should examine 543 hand preference in a concurrent, bimanual reaching-for-food task. In one scenario, 544 the manual operations should require a power grip followed by a precision grip; in 545 another scenario, the manual operations should require a precision grip followed by 546 a power grip. Contrasting hand-usage patterns in these two scenarios would 547 indicate that the individuals preferentially used the two hands depending on the 548 requirements of the tasks, that is, one hand to perform the manual operations 549 involving maneuvering in three-dimensional space and the other hand to perform 550 those involving physical strength. One should then examine hand performance with 551 regard to the requirements of the tasks in a concurrent, bimanual hand-552 performance-differentiation task. In one scenario, this task should ergonomically 553 force the usage of either the left or the right hand to perform a manual operation 554 requiring either a power grip or a precision grip; in another scenario, this task 555 should ergonomically force the usage of either the left or the right hand to perform 556 a manual operation requiring a precision grip and the other hand to perform the 557 one requiring a power grip. A more effective and/or efficient power grip in one 558 scenario and a precision grip in the other scenario would indicate that the 559 individuals used the two hands depending on the specializations, that is, difference 560 in the manual dexterity of the two hands. 561

Performance in Terms of the Efficiency of the Power and Precision Grips 564
Manual asymmetries might have ecological disadvantages as they can potentially 565 make an individual vulnerable to attack/defend appropriately only when the 566 prey/predator is present on a particular side. Also, as the stimuli are randomly 567 located with respect to the sagittal plane of an individual, i.e., towards left or 568 towards right, it might make it difficult to solve a particular task. However, manual 569 asymmetries are likely to help increasing manual specialization, the benefits of 570 which surpass the associated ecological disadvantages (reviewed by Vallortigara 571 and Rogers [2005]). Trehub [1983] drew a distinction between mere hand 572 preference and manual specialization by exemplifying human infants who exhibit 573 manual specialization and not hand preference (this idea was carried forward by 574 Fagot and Vauclair [1991] in non-human primates). According to Trehub [1983], 575 hand preference refers to the consistent usage of one hand to solve familiar, 576 relatively simple, and highly practiced tasks, and may not be necessarily 577 accompanied by an improvement in hand performance; whereas manual 578 specialization refers to the consistent usage of one hand to solve novel, relatively 579 complex, and not-practiced tasks that require peculiar action patterns, and is 580 necessarily accompanied by an improvement in hand performance. Trehub [1983] 581 also described that individuals generally exhibit manual specialization only in the 582 context of tasks that involve cognitively demanding manual actions (see, for 583 example, Mangalam et al. [2014b] that showed manual specialization in bonnet 584 macaques in tasks requiring peculiar action patterns viz., in terms of tasks that

Conclusions 633
Studies have investigated the evolutionary origin of hand-preference in non-human 634 primates. A careful analysis points towards the division of labor as being a general 635 principle underlying manual asymmetries. This principle is based on the difference 636 in the intrinsic requirements of the tasks, which can be broadly divided into 637 maneuvering in three-dimensional space and providing physical support, acquiring 638 power and precision grips respectively. Our review of studies on hand-usage 639 patterns in non-human primates reveals conceptual and logistic problems with the 640 spontaneous and experimental tasks used to determine hand-usage patterns; 641 moreover, methodology differs and confounding variables are often not 642 appropriately addressed. We suggest that studies on manual asymmetries in non-643 human primates should design experiments that do not undermine this possibility. 644 As far as the adaptive value of manual asymmetries are concerned, we suggest 645 that, to obtain more unambiguous answers, studies should be conducted with 646 experimental designs that allow comparing hand-usage patterns across species that 647 vary in their phylogenetic relatedness and/or ecology, over a range of spontaneous 648 activities and experimental tasks. It might be useful to study manual preferences 649 not just in isolation, but within the broader scope of the behavioral repertoire of the 650 species. Also, it might be advantageous to study the ontogeny of manual 651 preferences. Studies of these kinds may help to understand the forms and functions 652 of manual asymmetries, and the potential selection pressures under which manual 653 asymmetries are likely to appear and evolve. 654