The research team, led by Carina Spengler and Simon Saurbier at the Karlsruhe Institute of Technology together with colleagues including Andreas Lindenmann, Dieter Krause, and Sven Matthiesen, tackled a stubborn gap in the science of hand-arm vibration. Prolonged exposure to tool vibration is a well-documented occupational hazard, linked to vascular, neurological, and musculoskeletal disorders, including the painful condition known as white finger syndrome. The international standard ISO 5349-1 governs how this exposure is measured, and a companion standard, ISO 10068:2012, provides reference values for the mechanical impedance of the hand-arm system—the ratio of dynamic force to vibration velocity at the point where the hand grips the handle. But those reference values were built on limited grip and push force ranges, and earlier studies typically tested only a handful of force combinations with small groups of subjects.

What makes the new work stand out is its systematic scope. Rather than testing two or three discrete force settings, the team measured the biodynamic response across a full ten-by-ten grid of grip and push forces, with grip forces spanning 10 to 135 newtons and push forces spanning 10 to 110 newtons. Five subjects—three men and two women aged 23 to 34—each completed 300 measurements over multiple sessions spread across roughly twelve days, with the order of force combinations randomized to guard against fatigue and learning effects. In total, 1500 measurements were recorded on a test rig equipped with two electromechanical shakers, one axial and one rotational, driving a standardized cylindrical measuring handle instrumented with one-axial and three-axial force sensors and accelerometers.

The excitation itself was engineered for efficiency and safety. Instead of sweeping through frequencies one at a time, the shaker delivered a multisine signal—a superposition of sine waves from 10 to 500 hertz in 5-hertz steps—which compressed each measurement to just 16.5 seconds. The unweighted root-mean-square acceleration was 30 meters per second squared, far higher than the 2.5 to 9 meters per second squared typical of earlier single-sine studies, and the calculated total vibration value of 6.5 meters per second squared per measurement remained well below the exposure limits that would trigger protective measures. Subjects watched a real-time graphical display of their applied grip and push forces against the targets, adjusting their effort as needed, and any measurement deviating more than 5 newtons or 14 percent from the target was discarded—150 of the 1500 recordings failed this test.

Two quantities anchored the analysis. The mechanical impedance, expressed in newton-seconds per meter, describes how the hand-arm system dynamically resists motion: its real part reflects energy dissipated per unit time, while its imaginary part captures energy that is reversibly stored and returned. The vibration power absorption, measured in watts, is derived from the real part of the impedance and the squared vibration velocity, and quantifies the actual mechanical power flowing into the body. Both were evaluated along the translational z-axis of the hand-arm system, the direction defined in ISO 5349-1 as the primary axis of exposure assessment.

The results reveal a striking frequency-dependent choreography. Within a single subject, varying the grip and push forces broadened the median impedance magnitude from values between roughly 100 and 430 newton-seconds per meter into a range spanning 50 to 950 newton-seconds per meter across the measured frequencies; for one subject the spread reached a 15.8-fold increase at 50 hertz. Grip force proved the dominant parameter, strongly influencing impedance between 30 and 70 hertz and again above 160 hertz, with a maximum difference of 318 newton-seconds per meter at 40 hertz and a difference of 290 newton-seconds per meter persisting even at 500 hertz. Push force mattered most in the mid-range between 50 and 110 hertz, peaking at 177 newton-seconds per meter at 55 hertz, but its influence faded above 160 hertz—consistent with the fact that grip is generated by the hand and forearm while push originates from the shoulder, and high-frequency vibrations barely propagate beyond the hand itself.

The power absorption data told an equally nuanced story. Most of the absorbed energy concentrated below 60 hertz, precisely the range in which hand-held tools transmit the most vibration to the body, and the curves peaked between 35 and 40 hertz before collapsing toward zero above 100 hertz. Higher forces generally increased absorption, but in a narrow low-frequency window—roughly 10 to 30 hertz—the trend reversed, with stronger grips actually reducing the energy absorbed. The researchers attribute this to stiffening: as muscle contraction increases, the hand-arm system becomes a harder spring, reflecting more energy rather than dissipating it, while simultaneously improving mechanical coupling that channels more vibration into the arm at higher frequencies. Notably, the two female subjects showed lower impedance values above 100 hertz and different resonance features than the male subjects, echoing earlier reports that anthropometry and sex-related differences shape the biodynamic response.

Perhaps the most practically consequential finding concerns prediction. Because the force-dependent curves proved approximately linear at nearly every frequency, the team tested whether intermediate force levels could be interpolated from a small number of measured reference curves. The answer was emphatic: with just three carefully spaced grid curves, linear interpolation predicted both impedance magnitude and power absorption at intermediate forces with a mean absolute percentage error below 5 percent, corresponding to roughly 95 percent accuracy. Adding a fourth curve improved matters further—dropping the mean error for impedance to 3.29 percent—but with diminishing returns, and the largest accuracy gain came from moving from two to three curves. For future experiments, this means the grueling full-grid protocol may be unnecessary: a handful of reference force levels, distributed across the range of interest, could capture the essential behavior while dramatically reducing the burden on human subjects.

The study also delivers a pointed critique of the existing standard. When the measured impedance curves were compared against the ISO 10068:2012 reference range, they frequently fell outside it, especially between 20 and 100 hertz and above 250 hertz—exactly the frequency bands where grip and push forces exert their strongest influence. The authors argue that interaction forces should therefore not be treated as a fixed boundary condition in vibration analysis but as an explicit modeling parameter, one that varies continuously during real tool use. Their data, published as an open dataset, and their interpolation framework pave the way for physical and simulation-based hand-arm models that incorporate user forces directly—enabling virtual testing of tools before prototypes exist, robotic replication of human dynamics, and ultimately the design of machines that shake less hard on the people who hold them.

Subject of Research: Effects of grip and push force variations on the mechanical impedance and vibration power absorption of the human hand-arm system

Article Title: Modeling the influence of grip and push force variations on the mechanical impedance and vibration power absorption of the hand-arm system

Article References: Spengler, C., Saurbier, S., Lindenmann, A., Doellken, M., Bunk, L., Krause, D., & Matthiesen, S. (2026). Modeling the influence of grip and push force variations on the mechanical impedance and vibration power absorption of the hand-arm system. Results in Engineering, 32, Article 113151. https://doi.org/10.1016/j.rineng.2026.113151

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113151

Keywords: hand-arm vibration, mechanical impedance, vibration power absorption, grip force, push force, biodynamics, power tools, occupational health, ISO 10068, vibration exposure, interpolation, human-machine systems

The research tackles a long-standing question in sensorimotor science. When we reach for an object, our grip force rises in anticipation of its weight, a phenomenon known as anticipatory grip force control. If the prediction is right, the lift feels smooth and effortless. If it is wrong, the object either slips from our fingers or jerks upward with comical force. Decades of work have shown that these predictions are built from two ingredients: visual cues, most obviously the apparent size or volume of the object, and the memory of recent lifting experience. A large-looking bottle is expected to be heavy; a bottle that felt heavy last time is expected to feel heavy again.

That second ingredient turns out to be surprisingly dominant. In the Durham experiments, when participants lifted objects with no informative visual cues at all, their grip force behaviour was strongly biased by whatever they had lifted on the previous trial. This serial dependence, the tendency of the most recent experience to colour the next prediction, has been documented in perception before, but the new results show it exerting a powerful pull on motor planning itself. Left to its own devices, the lifting system essentially assumes the world has not changed since the last interaction.

The team measured this bias using peak grip force rate, or PGFR, the speed at which grip force climbs during the initial phase of a lift. PGFR is a well-established window into the brain’s internal heaviness predictions because it reflects what the motor system expected before tactile information arrived. Participants performed a precision-grip lifting task in which objects came in three heaviness levels, and the researchers tracked how PGFR shifted across four experimental conditions: a no-cue baseline, a familiar cue, and two kinds of novel cues.

The familiar cue was visual volume, the everyday signal that bigger means heavier. The novel cues were deliberately arbitrary: patterns of line orientations printed on the surfaces of the objects, with no natural relationship to weight. Crucially, the researchers also split the novel cues by presentation format. In the intrinsic condition, the line patterns appeared physically on the objects themselves, much as a label might. In the extrinsic condition, participants saw diagrammatic depictions of the same objects and patterns on a screen, meaning the weight information was available only through an interpretive step rather than being directly attached to the thing being lifted.

Before running the study, the authors had reasoned that the familiar volume cue might carry more predictive weight than an arbitrary pattern, simply because a lifetime of experience has calibrated the size-weight relationship. They also suspected that cues printed on the object might outperform cues shown on a screen, since intrinsic presentation requires no mapping between a display and the physical world. Neither prediction survived contact with the data. The familiar volume cue did not exert a stronger influence than the novel cues, and the intrinsic presentation offered no measurable advantage over the extrinsic one.

What did matter was simply having a cue at all. All three visual cue conditions significantly reduced the serial dependence bias compared with the no-cue baseline. In other words, any reliable visual signal about weight, whether natural or arbitrary, on-object or on-screen, was enough to loosen the grip of recent experience and let the motor system plan each lift on its own terms. The computational modelling accompanying the behavioural data, with individual cue-weighting parameters fitted for each participant, supported the same conclusion: novel cues were integrated into the heaviness prediction with weights comparable to those of the familiar cue.

The speed of this learning is the most striking aspect of the findings. Participants did not need weeks of training to make an arbitrary line orientation meaningful. The sensorimotor system, often portrayed as conservative and heavily reliant on long-term calibration, absorbed the new cue rapidly and deployed it in service of grip force planning without any apparent performance penalty. This suggests that the machinery of motor prediction is less wedded to specific cue types than theories emphasising learned cue familiarity might imply, and more open to whatever statistical regularities the environment currently offers.

The equivalence of intrinsic and extrinsic presentation carries equally interesting implications. Much of the emerging field of sensory augmentation assumes that information delivered through artificial channels, such as a display or a wearable device, will be harder to incorporate than information arriving through the body’s native senses. These results suggest that, at least for weight prediction in a controlled lifting task, the brain treats a diagrammatic on-screen cue as a perfectly serviceable predictor. The interpretive distance between a screen and an object did not dilute the cue’s influence on motor planning.

The practical implications extend beyond the laboratory. In industrial settings, medical environments, logistics and teleoperation, workers routinely handle objects whose weight cannot be judged by appearance, from sealed containers to virtual or robotic proxies. The findings suggest that equipping such environments with arbitrary visual markers, even simple line patterns shown on a monitor, could rapidly improve the safety and efficiency of manual handling by steadying grip force against the misleading pull of recent experience. For designers of augmented reality interfaces, the message is encouraging: the human motor system appears ready to trust whatever weight information you show it, almost immediately, and without demanding that the information be painted onto the object itself. As the authors note, the work highlights the flexibility of sensorimotor control and may inform manual interactions in environments with limited or unconventional weight predictors, a description that fits an increasing share of the technological world we now lift, carry and manipulate every day.

Subject of Research: How novel visual cues are learned to update internal predictions of object weight during lifting

Article Title: Learning a novel visual cue to update internal predictions of object weight

Article References: Learning a novel visual cue to update internal predictions of object weight. (n.d.). https://doi.org/10.1186/s12915-026-02741-1

Image Credits: AI Generated

DOI: 10.1186/s12915-026-02741-1

Keywords: grip force, motor planning, visual cues, object weight, serial dependence, sensorimotor control, anticipatory control, precision grip, sensory augmentation, psychophysics, BMC Biology, Durham University