NEURAL AND BIOMECHANICAL SPECIALIZATIONS OF HUMAN THUMB MUSCLES REVEALED BY MATCHING WEIGHTS AND GRASPING OBJECTS

1. Human manual dexterity has been linked by some to biomechanical ada ptations of the hand and by others to neural adaptations. To investiga te neural adaptations, the present study using the performance of four muscles acting on the index and thumb, quantified weight matching and electromyography. 2. The accuracy (i.e. reproducibility) of weight ma tching was used to investigate whether thumb muscles (i.e. flexor poll icis longus (FPL) and adductor pollicis (AP)) perform differently from index muscles (i.e. flexor digitorum profundus (FDP) and first dorsal interosseous (FDI)), and whether intrinsic hand muscles (AP and FDI) perform differently from extrinsic ones (FPL and FDP). 3. Subjects lif ted reference weights on the right which represented predetermined per centages of a force generated in a maximum voluntary contraction (MVC) ranging from 2.5 % to 35 % MVC (and to 50 % MVC in two muscles) and m atched them with a variable weight lifted in the same way on the left. 4. Analysis of the coefficients of variation (c.v., expressed as a pe rcentage) and the standard deviations calculated for repeated estimate s of perceived heaviness, revealed significant differences in the accu racy of weight matching between different muscles and between referenc e weights. Based on the c.v., subjects lifted more accurately with FPL and AP (the two thumb muscles) than with the two index muscles. The t wo intrinsic hand muscles (FDI and AP) were equally accurate, and sign ificantly more accurate than FDP which was the least accurate muscle. The high accuracy for FPL remained when accuracy was expressed in term s of the torque produced by the muscles when lifting the reference wei ghts, and also when the torques were converted to absolute intramuscul ar forces. 5. Accuracy (based on c.v.) decreased significantly with li ght weights and increased with heavy weights for all muscles except FP L, which was equally accurate over a very wide range of weights (< 2.5 % to 50 % MVC). When data from all muscles were pooled, the c.v. incr eased from 12.9 to 19.1 as the weights lifted decreased from 35 % to 2 .5 % MVC. 6. To examine the functional implications of the weight-matc hing study, electromyographic activity (EMG) was recorded with fine-wi re electrodes from the same four muscles while subjects lifted cylinde rs of different widths (17-50 mm) and weights (15-1000 g). For recordi ngs in which integrated EMG was linearly related to force up to maxima l levels, the amplitudes of the EMGs at 'lift off' and at the mid-poin t of the 'hold' phase of the task were expressed relative to the maxim al EMG during a MVC. Analysis of the amplitudes showed significant dif ferences related to the phase of the task and the width and weight of the cylinder. The relative EMGs increased linearly as the weights gras ped increased. 7. From the weight-matching study, we determined the fo rce above which the c.v. for weight estimation was below 16 % and used this to define an 'accurate' range of muscle force. The relative EMGs developed by the two thumb muscles while grasping and lifting objects that exceeded 50 g corresponded to forces within the 'accurate' range while those of the index finger were in the 'accurate' range only whe n lifting and grasping objects that exceeded 500 g. Based upon this cr iterion, these two index muscles typically act within the 'inaccurate' range during grasping. Thus, there may be particular reliance upon th e thumb muscles for precise control of forces. 8. The EMGs during gras ping also depended upon how the cylinder was lifted, i.e. pinch grip v ersus attempted encirclement of the cylinder by the index and thumb. W hen a very wide cylinder (75 mm diameter) was encircled by the thumb a nd index and lifted, the EMGs of all muscles (especially FPL and AP) w ere significantly less than those when the cylinder was grasped with a pinch grip. This biomechanically efficient mode of grasping is relate d to the evolution of a long and opposable thumb. 9. When data on perc eived heaviness from all muscles and weights were combined, we found a linear correlation between the standard deviations of the estimates a nd the reference weights when both were expressed in absolute terms (r = 0.99). It would appear that the central nervous system has the capa city to extract signals about weight independent of the torque or intr amuscular forces produced by the muscles. Except for small weights lif ted by FDP, this task is performed similarly for a range of muscles in the upper limb. 10. While both afferent and efferent mechanisms may c ontribute to the accuracy of weight matching, the findings are consist ent with the view that the neural control of the two human thumb muscl es studied has become more specialized than control of those acting on the index finger. This specialization may be related to the wide rang e of absolute forces that the thumb muscles must oppose in daily tasks .