Octopus arms exhibit exceptional flexibility

Octopuses are soft-bodied molluscs with eight sucker-lined arms that function in diverse behaviors such as locomotion, grasping, manipulation, and sensing. Octopus arms are muscular hydrostats—structures lacking rigid skeletal elements that rely on incompressible muscle tissue and internal pressure to generate support and movement. Because muscles are arranged longitudinally, transversely, and obliquely, and operate under a constant-volume constraint, such systems have theoretically high freedom for localized, complex movements. Prior literature often assumes octopus arms have exceptional flexibility and virtually unlimited degrees of freedom, but comprehensive measurement across all arm portions is lacking. Muscle architecture suggests that octopus arms can perform bending, torsion, elongation, and shortening, yet potential structural or neural control differences along the arm or among arm pairs might impose constraints. The present study examines how diverse arm deformations are in Octopus bimaculoides, asking whether all eight arms and all arm regions (proximal, medial, distal) exhibit all four deformation types and in which directions. By systematically recording deformation types across arm regions in laboratory-elicited behaviors, the study aims to empirically evaluate the long-standing assumption of exceptional arm flexibility and provide a baseline for future biomechanical and neurobiological analyses.

Background literature establishes octopus arms as muscular hydrostats capable of complex, localized movements due to the arrangement of longitudinal, transverse, and oblique muscles and associated connective tissues. Prior work documents arm functions (locomotion, manipulation, sensing) and motor programs (reaching via bend propagation with stiffening and elongation; fetching via pseudo-joint strategies). Studies in Octopus vulgaris indicate differences in muscle group density, orientation, and connective tissue interactions that may functionally constrain biomechanics despite similar physiological properties, and report both right- and left-handed helical muscle groups enabling bidirectional torsion. Behavioral studies suggest anterior arms are often preferred for exploration and manipulation, with possible lateral biases and posterior arm roles in locomotion. Despite this, comprehensive, region-by-region quantification of deformation types across all arms has been limited, motivating the current observational quantification in O. bimaculoides.

Subjects: Ten wild-caught Octopus bimaculoides (7 males, 3 females; ~400–700 g) collected off southern California and housed individually at the Marine Biological Laboratory (Woods Hole, MA) with environmental enrichment and fed live prey. Elicitation and recording: Animals were video-recorded over many months in several tank setups to elicit diverse arm movements: (1) open tanks without barriers; (2) tanks with plexiglass barriers containing a single hole permitting one or two arms; (3) a tank with a barrier containing many holes. Objects or food items on rods (e.g., plastic jack, clip, textured cylinder, bulb, rubber stopper) were presented to attract attention and provoke arm use. One or two cameras recorded from various angles. Footage was trimmed to remove inactivity or obstruction. Scoring: For all eight arms, observers recorded all fully visible occurrences of four deformation types: (a) bending (vertical and lateral planes; directions coded as oral, aboral, inward, outward), (b) torsion (clockwise, counter-clockwise), (c) elongation, and (d) shortening. Elongation/shortening were judged visually via changes in sucker spacing and arm thickness (constant-volume muscular hydrostat behavior). Bending angles were not measured; directions were assigned to the nearest of four categories. Each deformation’s location was coded as proximal, medial, or distal third of the arm. Sucker dependence was noted (sucker-dependent vs sucker-independent, the latter occurring without suckers contacting a surface). Stiffening was not scored due to feasibility limits. Observers and reliability: Three trained observers (not involved in filming) jointly reviewed footage, established scoring agreement before data collection, and periodically reconciled grading. Inter-observer reliability was assessed on eight clips using Pearson correlation coefficients averaged with Fisher z′ transforms; average r = 0.842, indicating acceptable reliability. Statistics: Counts of deformations were compared across arm pairs and regions using Pearson Chi-square tests with Bonferroni adjustments; post hoc comparisons used adjusted residuals. Directional torsion comparisons and lateralization were also evaluated with Chi-square tests. Animal use protocol was not required (invertebrates are not regulated in USA). Total analyzed footage: 120 minutes.

• Overall observations: 16,563 arm deformation events recorded in 120 minutes. All four deformation types (bending, torsion, elongation, shortening) were observed in all eight arms and in proximal, medial, and distal regions.
• Bending: Most common deformation, totaling 11,074 bends. Anterior arms (pairs 1 and 2) exhibited the majority of bends (8,227) versus posterior arms (2,847); χ² = 252.56, p = 0.006. Oral bending was least frequent among bending directions (oral = 1,347 of 11,074; χ² = 19.42, p = 0.001). Bending constituted the greatest proportion of deformations in medial (5,489 of 7,775) and distal (4,121 of 5,251) regions (χ² = 4680.06, p = 0.001), and was comparatively less frequent in the proximal region (1,464 of 3,537).
• Elongation and shortening: Elongation observed 2,723 times, more often in anterior than posterior arms (1,887 vs 836; χ² = 112.40, p = 0.003). Shortening observed 1,340 times, with no significant anterior–posterior difference (735 vs 605; χ² = 0.34, p = 0.55). Both elongation and shortening comprised a greater proportion of deformations in the proximal arm region than in medial/distal regions (χ² = 4680.06, p = 0.002).
• Torsion: 1,426 torsions recorded; no significant difference between clockwise (739) and counter-clockwise (687) directions overall or within regions (proximal CW 161 vs CCW 145; medial CW 346 vs CCW 362; distal CW 232 vs CCW 180; χ² = 5.53, p = 0.06). No anterior–posterior difference (χ² = 0.74, p = 0.39). Proximal torsion was rare in the first arm pair (15 occurrences out of 1,426).
• Lateralization: All deformation types occurred more frequently on the right than left side (χ² = 10.49, p = 0.015), except for the hectocotylized arm in males (R3), which showed a different pattern (p = 0.001).
• Sucker dependence: All deformation types occurred more often without sucker contact (sucker-independent) than with sucker use (χ² = 127.41, p = 0.008), indicating suckers are not essential for executing deformations.
• Deformation combinatorics: Considering 192 possible local combinations (four bending directions, two torsion directions, elongation, shortening across three regions and eight arms), 191 were observed at least once; only proximal counter-clockwise torsion in R1 was not recorded.

The study empirically validates the long-held assumption that octopus arms are exceptionally flexible, capable of producing bending, torsion, elongation, and shortening across all arms and regions. While all combinations appear possible, relative usage differs by region and task context: bending predominates overall, especially in medial and distal regions, whereas proximal segments more frequently undergo length changes (elongation/shortening). Oral bending was least frequent, likely reflecting experimental constraints and the functional tendency to orient the oral sucker surface towards targets. The findings align with muscular hydrostat biomechanics: coordinated activation of longitudinal, transverse, and oblique muscle groups, with bidirectional torsion supported by right- and left-handed oblique layers. Sucker use can enhance maneuverability but is not required to execute deformations. Behaviorally, anterior arms are preferentially used for exploration and manipulation, consistent with previous reports and suggesting task-related arm specialization and potential lateral biases. Neural control considerations indicate that simplifying strategies (e.g., bend propagation with stiffening during reaching; pseudo-joint fetching) underlie complex arm kinematics, and the observed regional differences (proximal elongation, medial bending) mirror these motor plans. The documented breadth of deformations and their distribution provides a foundation for refining models of octopus motor control and for informing bio-inspired soft robotic design where multifunctional, highly deformable appendages are desired.

This work provides comprehensive observational evidence that all eight arms and all regions of Octopus bimaculoides can execute the principal deformation modes—bending in multiple directions, bidirectional torsion, elongation, and shortening—demonstrating exceptional flexibility. Quantitatively, bending dominates overall usage, with regional and arm-pair differences reflecting functional roles and task context. The results supply a broad baseline for biomechanics and neurocontrol studies and offer actionable insights for soft robotics inspired by octopus arms. Future research should: (1) quantify stiffness as a fifth deformation and integrate 3D kinematics and precise angle/length measurements; (2) examine finer-grained regional subdivisions (e.g., by sucker sets) and simultaneous multi-arm tracking; (3) test more challenging manipulation tasks to probe limits; (4) extend to field observations and across life stages (including hatchlings) and species comparisons; and (5) integrate anatomical and neurophysiological measurements to link structure, control, and function for bioengineering applications.

• Observational, laboratory-based design prioritized eliciting diverse movements rather than representing natural behavior; barrier configurations favored frontal/anterior arm use.
• Bending directions were constrained to two orthogonal planes (vertical, lateral), not capturing intermediate angles.
• Stiffening, an important deformation, could not be assessed from video.
• Filming did not allow precise, simultaneous quantitative measurement of all eight arms or detailed 3D reconstructions.
• Visual assessment of elongation/shortening lacked precise metric quantification; bending angles were approximated to nearest category.
• Despite observer training, observational scoring is subject to human judgment, though inter-observer reliability was acceptable (r = 0.842).