The Comprehensive Core Training Guide

When people hear the term “core muscles” in common usage, they often think of the abs and obliques. These are indeed core muscles, but the core musculature is comprised of so much more, and deepening your understanding of this topic may enhance your ability to reach strength and physique goals. Technically, many of the muscles that function exclusively at the hip joint, such as the gluteus medius, are classified as core muscles (20). However, I already discussed many of these hip muscles on Stronger By Science, so we will be specifically exploring the core’s trunk muscles that have lumbar and thoracic spinal functions.

The Trunk

The thoracic spine is comprised of 12 vertebrae that constitute the skeletal upper and middle back, while the lumbar spine is comprised of 5 vertebrae that constitute the skeletal lower back. Spinal movements can occur in all three planes of motion at the joints that are formed between adjacent vertebrae (i.e., intervertebral joints). 

In the sagittal plane, flexion (i.e., rounding forwards) and extension (i.e., arching backwards) can occur. In the frontal plane, lateral flexion (i.e., side bending) can occur to the right or left. In the transverse plane, rotation (i.e., twisting) can occur to one side or the other. Compared to the lumbar intervertebral joints, the thoracic intervertebral joints have much greater transverse plane available ROM (range of motion) but lower sagittal plane ROM (78).

While the thoracic spine can move while the lumbar spine mostly maintains a fixed position and vice versa, I will refer to movement occurring at both regions as a trunk movement for the sake of simplicity. Some muscles exclusively act on the intervertebral joints in one region and different trunk exercises may bias the muscles in one region more than another, but typically an exercise that trains one region can also be performed in a manner to train the other. With respect to a particular core muscle’s ability to generate torque, variation can also exist across the multiple joints within the same spinal region. For example, the psoas major has greater leverage for laterally flexing the lower lumbar spine than the upper lumbar spine (41, 47). If you wish to delve into the finer details of this variation, you can read through the biomechanical studies I cite, but I will not discuss this degree of nuance in this article because it is not really pertinent to practical application. In the context of this article, I will be operationally using the term “trunk” exclusively to describe the lumbar and thoracic spinal regions, which differs from how other people may use this term. Some researchers define trunk motion as “a compound movement involving the simultaneous rotation of the lumbar spine, pelvis, and hips,” which would result in hip extensor muscles such as the gluteus maximus and biarticular hamstrings being classified as trunk extensors (91).

Of particular importance for understanding the functions of the core musculature is the fact that the right and left sides of a core muscle pair can act on opposite sides of the same intervertebral joint. This can result in the right and left sides of a muscle pair functioning as antagonists (i.e., opposing each other) in the frontal and/or transverse plane while acting unilaterally but functioning as synergists (i.e., contributing to the same function) in the sagittal plane while acting bilaterally. To help clarify this concept, let’s compare the internal obliques to the biceps in a simplified example that hypothetically assumes no other muscles are engaged beyond the muscles that are mentioned. When your right biceps contract while your right triceps are relaxed, your right elbow can flex regardless of how much your left biceps are contracting because your left and right biceps are acting on different elbow joints. In this instance, the origin and insertion of the right biceps are pulled closer together, while the origin and insertion of the right triceps (i.e., the biceps’ antagonist) move further apart. However, if your right and left internal obliques generate the same magnitude of force while contracting, the spine will not laterally flex or rotate because the equal and opposite forces counteract each other in those planes. Rather, bilateral internal oblique contraction will cause the spine to flex forward in the sagittal plane, which causes the origins and insertions of both sides of the internal obliques to move closer together. If your right internal oblique contracts while your left internal oblique is relaxed, the spine can laterally flex to the right or rotate to the right, both of which are actions that cause the right internal oblique to shorten while the left oblique (i.e., its antagonist) lengthens.

Let’s take a look at which muscles meaningfully contribute to the four different trunk actions and examples of exercises that train them. For each of these actions, you can perform either a dynamic or isometric (i.e., static) exercise. While you perform the trained function as a movement against resistance during the concentric phase of a dynamic exercise, you will resist movement that opposes the trained function during an isometric exercise. Especially with direct core training, no universally accepted technique exists for every exercise, and more than one name may be used for the same exercise, which can provide some ambiguity when using certain terms. The “back extension” exercise, which is also known as a “hyper extension” and can be performed on a 45° Roman chair or a glute ham developer, is a prime example of this. One lifter may situate his pelvis above the pad and dynamically flex and extend his hips while keeping his trunk in a fixed position. Another lifter may situate his pelvis below the pad and dynamically flex and extend his trunk while no motion occurs at the hip joint. People may refer to both of these lifts as “back extensions,” but the former exercise trains the trunk extensors isometrically and the hip extensors dynamically, while the latter exercise trains the trunk extensors dynamically and the hip extensors isometrically. As I mentioned in a previous Stronger By Science article, similar variation is present with how people perform movements such as hanging leg raises or reverse crunches. Depending on your technique, you can perform these exercises in a manner that trains your abdominal muscles dynamically or isometrically.

Trunk Flexion

Trunk flexion is primarily produced by the rectus abdominis, internal obliques, and external obliques, while the transversus abdominis can also assist to a lesser degree (13, 35, 41, 47, 69, 79).  Of these four muscles, which are all located within the anterior core, the rectus abdominis has the greatest leverage for generating trunk flexion torque. 

Dynamic trunk flexion exercises include variations of crunches, sit ups, hanging leg raises, and v-ups. Isometric trunk flexion exercises, where you resist extension, include front planks, ab wheel or barbell rollouts, “six inches,” and any exercise where bilateral hip flexion is performed dynamically while the trunk maintains a fixed position. 

Trunk Extension

Trunk extension is primarily produced by the intrinsic back muscles and secondarily assisted by the quadratus lumborum (AKA QL) and latissimus dorsi, all of which are located within the posterior core (13, 35, 41, 47, 59, 69, 79). The intrinsic back muscles run along the entire length of the spine, and two of its three layers act on the trunk. The intermediate layer contains the prominent erector spinae, which is a muscle group consisting of the iliocostalis, longissimus, and spinalis. In contrast to the iliocostalis and longissimus that can contribute to movement at the cervical, thoracic, and lumbar spine, the spinalis runs along the cervical and thoracic spine but not the lumbar spine. The deep layer contains the transversospinales, which is a muscle group comprised of the semispinalis, multifidus, and rotatores. While the multifidus and rotatores cover all spinal regions, the semispinalis stretches only across the cervical and thoracic vertebrae. 

Dynamic trunk extension exercises include reverse hypers, “supermans,” Jefferson curls, and some variations of “back extensions.” Isometric trunk extension exercises, where you resist flexion, include variations of squats, deadlifts, good mornings, kettlebell swings, cleans, snatches, and bent over rows. 

Trunk Lateral Flexion

Trunk lateral flexion is primarily produced by the internal obliques, external obliques, quadratus lumborum, transverse abdominis, and lats, while the erector spinae, rectus abdominis, and the psoas major secondarily assist (13, 35, 41, 47, 69, 79). 

Dynamic trunk lateral flexion exercises are often performed as some variation of a “side bend” for which you can use a cable, dumbbell, kettlebell, or band as resistance. You can also utilize your bodyweight as resistance for this movement with a side plank variation performed in a manner that trains dynamic trunk lateral flexion. Isometric trunk lateral flexion exercises, where you resist lateral flexion to one side, include static side planks, waiter walks (i.e., single arm overhead carries), and suitcase lifts. The suitcase position refers to when a weight is held with one hand at your side and can be used for carries, deadlifts, rack pulls, and lunges to train isometric trunk lateral flexion and grip strength (if you don’t use straps) concurrently. These exercises will also engage lower body muscles such as the glutes, however, strength of your core muscles is much more likely to be the limiting factor in performance of suitcase lifts unless your lower body muscles are already meaningfully fatigued from prior lifts.

Trunk Rotation

Trunk rotation is primarily produced by the internal obliques and external obliques and secondarily assisted by the lats (80, 98). Anatomy textbooks may list trunk rotation as a function of the erector spinae and transversospinales, but it is presently unclear how much of a role they actually contribute to this movement (73). Some of the intrinsic back muscles that belong to these groups have been measured to be active during isometric trunk rotation, but they may engage to merely resist other spinal movements such as lateral flexion and flexion, which could otherwise occur in response to contraction of the obliques (66, 80, 82, 98). While the ability of core muscles to function in the sagittal and frontal planes has been well studied for each spinal region, research quantifying the capacity of core muscles to act in the transverse plane is much more limited for the trunk. The orientation of the muscle fibers that comprise the erector spinae and transversospinales indicate that their capacity to rotate the spine is negligible (54). These muscles may generate a very small amount of trunk rotation torque, but the magnitude is miniscule enough to be practically irrelevant (54). Given that I have yet to see convincing empirical data that demonstrates that the intrinsic back muscles can meaningfully contribute to the production of trunk rotation torque, I will not presently attribute this function to them, although new data may emerge in the future that indicates otherwise.

Dynamic trunk rotation exercises include Russian twists and variations of chops that can be loaded with cables or elastic bands. With chop variations, you can adjust the height of the cable or band to provide different directions of resistance that enable you to perform the exercise exclusively in the transverse plane (i.e., no upward or downward motion) or in a diagonal manner (i.e., high to low or low to high motion). Isometric trunk rotation exercise, where you resist rotation to one side, is commonly performed as a Pallof press with a cable or a band that is directed perpendicularly to the direction in which you press. Alternatively, you can train isometric trunk rotation with a unilateral horizontal press or row using a cable or band that is aligned in a manner that directly opposes the direction in which you press or row. 

A Brief Note on the Lats

Biomechanically, the lats have meaningful moment arms that enable them to generate torque for trunk movements, and EMG (electromyography) research indicates that the lats are active during trunk extension, lateral flexion, and rotation (65, 67, 80, 82, 107). However, the lats also function as primary shoulder adductors and shoulder extensors, and I strongly suspect that exercises that train these movements (e.g., pullups and chin-ups) would provide a notably greater stimulus to the lats than any type of trunk exercise (2, 9, 46, 85). While no longitudinal training interventions have investigated this matter, we do have other forms of evidence indicating that trunk exercises will predominantly train other core muscles such as the erector spinae during extension and the obliques during rotation (17, 18, 19, 98). Consequently, I do not consider any trunk exercises to be a viable substitute for various vertical pulling exercises or certain types of rows with respect to training the lats.

How the Core Functions while Resistance Training

During various sport maneuvers such as swinging a golf club or a baseball bat, high velocity trunk movement is an important component of high performance (24, 42, 100). However, during heavy resistance training, individuals typically strive to use their core muscles to resist motion at the trunk’s intervertebral joints. A notable exception is when some powerlifters deliberately initiate the deadlift with a flexed thoracic spine prior to extending this region to an upright position during the lockout. Relative to if no thoracic motion occurred, this technique can shorten the moment arm of the resistive torque acting at the hip joint and consequently decrease the demands imposed on the hip extensor muscles for a given weight (36). Other than this instance, few experienced strength athletes or coaches would advocate that a lifter intentionally performs a visibly high degree of trunk movement while training squat or hip hinge variations (e.g., deadlift, Romanian deadlift, good morning). Rather, the conventional guidance is to purposely minimize any other trunk motion during these exercises as movement occurs at the hip and knee joints. When using a symmetrically loaded barbell for these lifts, resisting sagittal plane trunk movement will be most challenging because resistive torque will be greatest in the sagittal plane as the external load imposes flexion torque at the lumbar and thoracic intervertebral joints. Given that the lumbar joints are situated perpendicularly further away from the barbell than the thoracic joints when the torso is in an inclined position, resistive flexion torque will be greater at the lumbar region when squatting and deadlifting. Even with proficient technique and submaximal loads, a greater than expected degree of lumbar spinal flexion may still occur when squatting and deadlifting as demonstrated by research conducted by Aasa et al (2019) and Edington (2017).

Aasa et al measured the ROM that occurred at the upper and lower lumbar spine as 24 powerlifters and weightlifters with a mean of eight years of strength training experience performed 3-rep sets of back squats and conventional deadlifts with 70% 1RM loads (1). With respect to sagittal plane movement, which was slightly greater when deadlifting than squatting, a 9.7-11.8° mean ROM occurred in the upper lumbar spine and a 18.1-21.7° mean ROM occurred in the lower lumbar spine. For reference, some data indicates that 60-65° is a normal amount of flexion ROM for the whole lumbar spine for young adults, but lumbar spinal ROM can meaningfully vary among individuals, particularly if they belong to different age groups (81, 104). Just as the lumbar intervertebral joints will experience greater resistive flexion torque than the thoracic intervertebral joints when squatting and deadlifting, so too will resistive flexion torque be greater in the lower lumbar joints than the upper lumbar joints during these lifts. This difference, stemming from the perpendicular distance between the barbell and these joints, likely contributes to the greatest motion occurring at the lower lumbar intervertebral joints.

Edington also provided insightful data through a well-designed study on 17 well-trained strength athletes that assessed lumbar spine kinematics and kinetics as the participants performed back squat and conventional deadlift singles with 85% of their 1RMs (21). Relative to the maximum lumbar flexion ROM that participants were able to achieve without external load during a standing trunk flexion test, on average the athletes reached 64% and 77% of peak lumbar flexion respectively when squatting and deadlifting. For context, these flexion ROM values were calculated relative to the lumbar spine angles the participants were measured to have while standing upright (a proxy for a neutral position). In this upright position, the lumbar spine is naturally lordotic (i.e., extended beyond 0°) so a meaningful portion of flexion ROM can occur while the intervertebral joints are still at extended angles. For instance, one participant’s lumbar spine was measured to be in 51° of extension when standing and 31° of flexion during the max trunk flexion test. Consequently, if he reaches 64% of his available flexion ROM, his peak angle of lumbar flexion would be just 1.5° of flexion beyond 0°. However, lumbar spinal flexion ROM and lordotic angles can vary widely among people, so the same percentage of lumbar flexion ROM that begins from someone’s upright lordotic position can manifest as quite different angles of peak lumbar flexion. In the same study, another participant’s lumbar spine was measured to be in 23° of extension when standing and 65° of flexion during the max trunk flexion test, so this athlete would reach a 33° angle of peak flexion during the squat if his lumbar spine flexed through 64% of its available ROM. When comparing these two lifters, we can see that reaching the same angle of lumbar flexion during an exercise can have different implications for two different individuals. What may be a moderately flexed position for one person can be an extremely flexed position for another person when accounting for individual differences in lumbar spine lordosis and available flexion ROM.

With a 50% time point corresponding to when the deadlift lockout is completed and the peak depth is reached during a squat, the angle of peak lumbar flexion was reached approximately 9% into the deadlift (i.e., right after breaking the floor) and 49% into the squat (i.e., in the hole). Just before and just after this position of peak lumbar flexion was reached respectively while deadlifting and squatting, compressive and shear lumbar forces acting on the lumbar spine peaked as well. It’s worth noting that the squat data we are covering reflect the average of the values obtained from the separately measured high bar squats and low bar squats, which you can learn more about from Greg Nuckols. With respect to the lumbar spine, the only significant difference between these two styles was that greater peak lumbar flexion occurred when low bar squatting (68% of max flexion ROM) than high bar squatting (60% of max flexion ROM). Overall, the fact that Aasa et al and Edington assessed submaximal sets (70-85% 1-RM often corresponds to approximately a 12-6-rep max) suggests that a moderate to moderately high degree of peak flexion at the lumbar spine is a normal characteristic of squatting and deadlifting moderately heavy loads (63,76). 

Lifting with Lumbar Spinal Flexion 

If you have some experience with lifting, at some point you’ve likely heard the recommendation to maintain a neutral lumbar spine position during these exercises as opposed to allowing this region to flex dynamically under load. In reality you can dynamically flex and extend the lumbar spine to some degree while simultaneously operating within the neutral zone. The neutral zone is the ROM where “spinal motion is produced with a minimal internal resistance” (84). Just as with available joint ROM, the width of the neutral zone varies among the different intervertebral joints and among different individuals, which adds complexity to this matter (113). Furthermore, there is little reason to believe that the neutral zone measured in a cadaver spine under a constant load will be the same as the neutral zone for a live human being while lifting. For these various reasons, categorically considering whether a person’s lumbar spine as a whole is in a neutral position during a particular activity is not necessarily a simple task. With this said, the lumbar spine is nearly certainly no longer within the neutral zone when squatting or deadlifting respectively with well greater than 50% of available lumbar spine flexion ROM. For reference, one study on cadaver spines reported the initial 23% of maximum lumbar flexion ROM to constitute the neutral zone, which is considerably less than what was measured in the aforementioned study on squats and deadlifts (21, 113). Beyond the neutral zone lies the elastic zone, which is the ROM where “spinal motion is produced against significant internal resistance,” which is passively generated by structures such as ligaments, joint capsules, intervertebral discs, and bony articulations (39, 84). As with any other component of the musculoskeletal system, it is not inherently bad to stress these spinal structures; however, injury can occur in any tissue if it is loaded in excess of its load tolerance. Deadlifting and squatting heavy weights substantially loads the lumbar spine through a combination of compressive and shear forces, and the posture in which the lumbar spine is loaded can affect where stress is distributed throughout the spine’s structures (6, 14, 30).  

When discussing the potential relationship between lifting with a flexed lumbar spine and low back pain, I have previously seen other people refer to a systematic review with a meta-analysis conducted by Saraceni et al (2020). The researchers sought to answer the question “Is there a relationship between lumbar spine flexion during lifting and low back pain?” and found 12 studies that met their inclusion criteria (92). Ultimately, Saraceni et al concluded that “there is currently no credible longitudinal or cross-sectional evidence to suggest that a more flexed lumbar spine during lifting is a risk factor for low back pain onset or persistence…” In the context as a lifter, you may read through the abstract and conclusion that repeatedly refer to lifting and naturally equate lifting to resistance training, particularly if someone mentions this review while talking about lumbar position while squatting and deadlifting. However, none of the studies included in the review examined participants who performed resistance training, and the loads lifted in these studies ranged from a 26.5lb (12kg) box to a pen. Furthermore, 11 of the included studies were cross-sectional, so they compared how people who already had low back pain lifted very light loads compared to people without low back pain. The only longitudinal study assessed lumbar spine kinematics in nursing students as they picked up a pen, pillow, and 11lb (5kg) box (72). We simply cannot interpret these findings to mean that lumbar posture while deadlifting and squatting high loads has no effect on someone’s risk of developing low back pain. 

To my knowledge, no experiments or prospective observational studies have investigated how consistently deadlifting and squatting with a very high or low degree of lumbar spinal flexion may affect a lifter’s risk of developing a low back injury or pain. In the absence of this research, I cannot assert with complete certainty that one method of lifting is inherently riskier than the other. Nonetheless, if this was the only level of evidence that we accepted to be credible enough to guide technique recommendations for different exercises, we would be unable to provide virtually any guidance at all. We have no direct data on how bench pressing with fully extended wrists affects a lifter’s probability of developing wrist pain or how squatting with knees caving completely inward to the extent they bang each other affects someone’s likelihood of developing knee pain. This absence of direct data does not preclude competent coaches from drawing on other forms of evidence to discourage bench pressing and squatting in these manners. Similarly, strength training professionals widely recommend that lifters usually avoid deadlifting and squatting with a highly flexed lumbar position for a reason (99). Ultimately, I coach field sport athletes and recreational lifters to attempt to minimize lumbar spinal flexion when deadlifting and squatting to the best of their abilities in order to err on the side of caution. Maintaining a neutral position may not actually be possible during a heavy lift, but consciously attempting to do so can help a lifter limit how much lumbar flexion unintentionally occurs. Some evidence indicates that the lumbar spine can tolerate a higher volume of loading before injury occurs in a neutral position compared to a moderately flexed position (27). However, I deem the avoidance of extreme lumbar flexion when deadlifting and squatting to have the greatest potential impact on the likelihood that the load tolerance of certain tissues is exceeded while training.  

During a heavy deadlift, the high magnitude of flexion torque acting on the lumbar intervertebral joints can enable these joints to flex to a greater degree than could otherwise be achieved during an unloaded stretch. When a person’s lumbar spine flexes beyond its unloaded ROM during a heavy deadlift, injury to a lumbar spinal muscle, ligament, or intervertebral disc’s posterior annulus may be more likely to occur. If extreme dynamic lumbar flexion occurs while deadlifting, some lumbar spinal extensor muscles may experience an elevated risk of incurring a strain injury as they are subjected to a considerable magnitude of eccentric loading in a very stretched position. Similarly, the supraspinous and interspinous ligaments, which are situated on the back of the spine, are loaded to the greatest degree in a fully flexed position and can be stressed to the extent that they sprain if the lumbar spine is excessively flexed under load (6).  Research on human cadaver spines has indicated that intervertebral disc prolapse can also occur if an extremely flexed intervertebral joint is subjected to a sufficiently high magnitude or volume of loading (4, 5). In cadaver spines from animals with similar spinal structures, researchers additionally found that high rate compressive forces were able to induce disc herniation in joints that were loaded in a fully flexed position but not in joints that were loaded in a neutral position (109). I recognize that spines obtained from cadavers will not necessarily respond to loading in the same manner as those found within live human beings who have the ability to adapt. Nevertheless, I believe that these findings are worth noting without overgeneralizing the conclusions that we can draw from them.

Regardless of technique, injury can occur in response to an excessively high volume and intensity of any athletic endeavor whether that be deadlifting, running, jumping, or throwing. However, the technique that is used for any activity can affect how likely a particular workload is to exceed a particular tissue’s load tolerance. Your lumbar spine will not spontaneously combust if it is subjected to loaded flexion, nor will resisting any noticeable lumbar flexion while lifting eliminate your risk of low back injury. Nonetheless, I believe that your training has much room for improvement if you find yourself consistently deadlifting with a cat back and squatting while being completely folded forward at the spine. For another perspective on the topic of lifting with a flexed lumbar spine, you can read “Should You Fear Lumbar Flexion?” by Sam Spinelli on Stronger By Science.

Core Function for Performance

Beyond potentially affecting the likelihood that some tissues are loaded in excess of their load tolerances, a lifter’s ability to stabilize the trunk is invaluable for high performance. An excessive degree of inadvertent trunk flexion may contribute to failing a lift that requires an upright trunk position to be locked out. When a very high degree of trunk flexion occurs while squatting, you can become pitched forward into a disadvantaged position, lose tightness, and struggle to complete the lift. While you’ll likely be able to deadlift with more trunk flexion than while squatting, the ability to extend a considerably flexed trunk can still be a limiting factor in deadlift performance. Whether you opt to intentionally initiate a deadlift with a moderately flexed thoracic spine or a more extended position, high core strength is essential for high performance.

With a symmetrically loaded barbell, core muscles that cannot generate any trunk extension torque may seem to be irrelevant while squatting and deadlifting. Activating the abdominal and oblique muscles may even appear to be detrimental to resisting trunk flexion because contraction of these muscles can generate spinal flexion torque. However, optimal spinal stabilization requires the anterior and posterior core muscles to operate synergistically in an isometric manner to produce a great deal of intra-abdominal pressure, which stiffens the trunk (8). Gas pressure is inversely proportional to the volume in which the gas is contained, so a reduction in intra-abdominal volume generated by simultaneous recruitment of the anterior and posterior core muscles increases intra-abdominal pressure. In common usage, stiffness often has a negative connotation, but trunk stiffness is critical to performance for anyone who performs activities that can transmit considerable loading through the spine such as squats and deadlifts. Stiffness quantifies the capacity to resist deformation in response to a force being applied, so a high degree of trunk stiffness can help an individual resist inadvertent spinal movement in any plane of motion. Consequently, core muscles that cannot generate spinal extension torque, such as the rectus abdominis and transversus abdominis, can still serve a role in resisting spinal flexion when a lifter properly uses them to brace. This may in part explain why isometric trunk flexion endurance strongly positively correlates with maximal relative deadlift strength (i.e., 1RM/bodyweight) (43). The influence that intra-abdominal pressure can have on resistance training performance is the primary reason why many lifters opt to wear a belt during axially loading exercises such as squats and deadlifts. Compared to training beltless, training with a belt that is used properly allows most people to generate greater intra-abdominal pressure during these exercises (38, 48, 49). This is likely the reason why many lifters who are familiar with the technique of bracing against a belt report that they can squat and deadlift higher 1RMs or complete more reps with a given load when wearing a belt. For a deeper dive into how wearing a belt can affect resistance training, I recommend reading “The Belt Bible” by Greg Nuckols.

Increasing intra-abdominal pressure has the potential to boost performance further by enhancing torque production elsewhere in the kinetic chain. Dr. Stuart McGill, a notable spine expert and clinician, has repeatedly stated that “proximal stiffness (meaning the lumbar spine and core) enhances distal athleticism,” and this principle is on full display while lifting heavy weights (64). Whenever a muscle contracts, it generates a force directed in a manner to pull its anatomical attachments (i.e., insertion and origin) closer together. In order to maximize how much force is directed to generating torque at one end, the other end must be fixed. For instance, the gluteus maximus originates on the pelvis and inserts onto the femur. Contraction of this muscle can pull the femur backwards via hip extension, which pulls the insertion closer to the origin, or it can tilt the pelvis backwards via posterior pelvic tilt, which pulls the origin closer to the insertion. For optimal deadlift and squat performance, you want to maximize how much force generated by gluteus maximus contraction is diverted to generating hip extension torque. To do so, the pelvis must be stabilized by muscles that can tilt the pelvis forward via anterior pelvic tilt such as the erector spinae. Tayashiki et al (2016) investigated this matter by assigning 20 recreationally active participants to a control group or a training group for an eight-week intervention comprised exclusively of abdominal bracing without external resistance (103). Specifically, the training group performed five sets of 10 reps three times per week, where each rep was comprised of a two-second maximal isometric abdominal brace. After eight weeks, the training group increased mean isometric hip extension and trunk extension strength respectively by 35% and 14%, while all of the control group’s performance metrics were unchanged. This strength increase was accompanied by a 37% increase in maximal intra-abdominal pressure that the training group could generate while bracing. Furthermore, the training group increased peak power by 16% during an exercise functionally equivalent to a concentric only box squat from parallel depth. 

Ultimately, Tayashiki et al’s findings demonstrate that core training may transfer to greater lifting performance if it enhances your bracing technique and facilitates a more efficient production of intra-abdominal pressure. What may appear to be a core strength deficiency during a squat or deadlift can actually arise from improper bracing, particularly for individuals with low levels of training experience. However, at a certain level of technical proficiency, the force production capacity of some core muscles can directly limit performance. With the great degree of resistive trunk flexion torque that may be present while squatting and deadlifting, strengthening the posterior core muscles, which generate trunk extension torque and intra-abdominal pressure simultaneously, will likely yield the greatest carryover. While recruitment of the anterior core muscles is necessary for optimal bracing, their activation will still produce undesirable trunk flexion torque. Ultimately, the magnitude of advantage produced by the intra-abdominal pressure they develop can be greater than the magnitude of disadvantage incurred from their trunk flexion torque, resulting in a net benefit. Nonetheless, increasing the force produced from these muscles beyond a relatively low threshold is unlikely to yield any improvement in squat or deadlift performance as suggested by EMG research. Regardless of which load is used for squats or deadlifts, activation of the obliques and abdominals has been measured to be consistently well below their maximum (i.e., mean EMG values not exceeding 25% of maximal voluntary isometric contraction EMG values) (11, 16, 83, 89, 90, 112). In contrast, EMG measurements have indicated that the intrinsic back muscles are much more engaged while squatting and deadlifting, which is congruent with what we would predict given the biomechanical demands of these lifts (11, 15, 16, 83, 89, 90, 112). These EMG studies most commonly assessed the lumbar erector spinae, but a couple of studies also assessed the lumbar multifidus and found it to be quite active while squatting and deadlifting as well (15, 83). To my knowledge, activation of the thoracic erector spinae and multifidus has not been measured while squatting, but I would expect these muscles to be highly active while squatting just as they are while deadlifting (15). 

Are Squats and Deadlifts Enough?

Given the importance of the posterior core musculature for squat and deadlift performance, one would intuitively think that squats and deadlifts are ideally suited to strengthen this region. Biomechanically, I see no reason as to how heavy squats and deadlifts cannot provide a viable stimulus to strengthen these muscles, but using them in the absence of direct core exercises may be insufficient to maximize their force output. Androulakis-Korakakis et al (2021) tested lumbar spinal extension isometric torque at 5-7 different joint angles for recreationally trained (i.e., at least two years of resistance training experience) men, noncompetitive powerlifters (Squat 1RM = 390lb and Deadlift 1RM = 450lb), and competitive powerlifters (Squat 1RM = 474lb and Deadlift 1RM = 511lb) (7). Much to my disbelief, lumbar spinal extension strength was quite similar among the three groups, and the recreationally trained group had the highest mean values by a slight insignificant difference despite having a significantly lower body mass than the other two groups (7). It’s important to not put too much weight on the findings of any single study, particularly one that is not designed to determine if a causal relationship may exist, and I wish the recreationally trained men were tested for their squat and deadlift 1RMs to provide a clearer comparison. However, causality can be inferred from randomized trials that measure lumbar extension strength before and after training interventions. 

Hammond et al (2019) assigned 14 males with at least six months of resistance training experience to either a back squat or barbell hip thrust group as both groups performed two sessions per week for four weeks (37). Each session was comprised of three sets performed until failure with 80% of the assigned exercise’s 1RM, resulting in 6-10 reps per set. While some people will refer to failure as leaving no reps in reserve, this study had its participants continue the set until they literally failed the concentric phase of a rep. Six sets per week is not a particularly high volume of training, but the lifters in the squat group definitely had a grueling month where they intentionally got buried 24 times, so I was surprised to see 100% adherence for the study. Following the intervention, the squat group increased squat 1RM by 6.5% and hip thrust 1RM by 12.5%, while the hip thrust group increased squat 1RM by 6.8% and hip thrust 1RM by 15.4%. Mean isometric lumbar extension strength increased by just 1.7% for the squat group and 2.4% respectively for the hip thrust group. Neither of these changes were greater than the typical error of the measurement. 

In another randomized controlled trial, Fisher et al (2013) assigned 36 young adult male subjects with at least two years of resistance training experience (including a deadlift variation) to 10 weeks of training the Romanian deadlift or a machine lumbar spinal extension exercise (22). Each group performed a single set of 8-12 reps to volitional failure through a full ROM once per week, and participants were able to use straps for the Romanian deadlift training and testing 1RMs. While both groups significantly increased their Romanian deadlift 1RMs (by 16.1% for the Romanian deadlift group and by 7.5% for the lumbar extension group), only the direct lumbar extension group significantly increased its isometric lumbar spinal extension strength (22). An important takeaway from Fisher et al’s study is that increasing isolated lumbar extension strength through dynamic lumbar extension exercise can transfer to improving performance on a heavy hip hinge with free weights. Throughout the 10-week intervention, the participants refrained from any other exercises that trained the low back, glutes, or hamstrings.  For trained individuals, neglecting the previously trained hip extensor muscles for this duration of time may result in a loss in hip extension strength, which could impair Romanian deadlift performance. Consequently, the 7.5% increase in Romanian deadlift 1RM that was experienced by the lumbar extension exercise group after only 10 total sets over 10 weeks is more notable in this context compared to if a similar improvement occurred for untrained subjects. Given that the lumbar spinal extensors serve a very similar function during squats, deadlifts, and Romanian deadlifts, I would hypothesize that dynamic lumbar extension exercise can transfer to enhancing squat and deadlift performance as well. Because other variables such as hip extension strength can be the limiting factor during these lifts, the degree to which increasing lumbar extension strength transfers to enhancing them will vary among individuals.        

The two training interventions conducted by Hammond et al and Fisher et al certainly provide data that is worth noting, but we must be careful not to overgeneralize their findings. It would be erroneous to claim that they demonstrate that exercises such as deadlift or squat variations that require the posterior core muscles to act isometrically to resist lumbar spinal flexion are incapable of meaningfully increasing lumbar extension strength. Rather, I interpret these findings to indicate that those types of exercises are not the most efficient way to develop isolated lumbar extension strength with regard to set volume allocation. I liken this to comparing conventional deadlifts to hack squats with respect to inducing quad growth. Both exercises engage the quads, but performing one set of hack squats to failure can provide a meaningfully greater hypertrophic stimulus to the quads than one set of deadlifts to failure. Deadlifting may be able to induce quad hypertrophy, but hack squats are a more efficient means of doing so in large part because they ensure that quad strength is the limiting factor during a set and can train the quads at long muscle lengths. Perhaps some lifters may experience similar quad growth whether they perform a low volume of hack squats or a high volume of deadlifts. However, the volume of deadlifts that may be required to induce meaningful quad hypertrophy in trained lifters can impart much more systemic fatigue, and no growth may be detected if only low volumes are used.        

Similarly, higher volume Romanian deadlift or longer duration squat interventions may have resulted in detectable increases in lumbar spinal extension strength. While Hammond et al and Fisher et al stated that their subjects had lifting experience and were consistently training at the time of recruitment, we know very few specific details as to what their prior training included, which can certainly affect how they respond to a new protocol. For an undefined period of time directly preceding the interventions, Hammond et al’s participants were training back squats and hip thrust at least once per week, and Fisher et al’s participants were not performing Romanian deadlifts or “specific lumbar exercises.” Other than this, we don’t know anything about key variables such as the types of exercises and volumes that the subjects were previously utilizing. This element of uncertainty and lack of standardization of prior programs is typically the trade-off that is presented by recruiting trained participants for training interventions rather than assessing untrained individuals.  

Like Hammond et al, Fisher et al did not measure the size of any muscles throughout the study, which could have provided clarity as to how much of the documented strength gains resulted from hypertrophy as opposed to neural adaptations. Nonetheless, I suspect that a dynamic trunk extension exercise can be better suited to induce posterior core muscle hypertrophy than a hip hinge or squat. Fisher et al did not measure lumbar spine joint angles as participants performed the Romanian deadlift; however, the researchers coached the lifters to utilize the conventional technique where a flat back position or lordotic lumbar curvature is maintained (26, 29). In contrast, the machine lumbar extension group dynamically trained through 72° of lumbar ROM that would load the lumbar extensor muscles at longer lengths than the Romanian deadlift. As previously discussed on Stronger By Science, the peak muscle length that is loaded during an exercise has a meaningful effect on the hypertrophic stimulus that it can provide. Training a muscle at long lengths has been measured to produce greater increases in size than training at short lengths, likely due to stretch-mediated hypertrophy (55, 70, 87, 93). 

A dynamic lumbar extension exercise that trains through a rather flexed lumbar position loads the posterior core muscles at longer lengths than a hip hinge variation. Correspondingly, it would be reasonable to predict that the former type of training could yield greater hypertrophy than the latter if volume is equated. In the initial stages of resistance training, strength gains can be predominantly attributed to neural adaptations such as improved motor unit recruitment, but increasing muscle size plays a larger role for further strength development as an individual becomes more experienced (25, 77). Consequently, the value of training the posterior core muscles at long lengths to optimize hypertrophy may be greatest for more experienced lifters who wish to maximize long-term increases in trunk extension strength. 

Isolated Lumbar Extension Training

Multiple other resistance training interventions have also indicated that isolated dynamic lumbar extension exercise can be a very effective means of increasing isometric lumbar extension strength (12, 23, 32, 33, 34, 61, 75, 88, 101, 105). While many studies have examined dynamic lumbar extension exercise interventions, I am unaware of any research that assessed strength adaptations after a thoracic extension training intervention. Given that extension strength in both the lumbar and thoracic regions is important for squatting and deadlifting, lifters can benefit from training the intrinsic back muscles in both regions. With the functional overlap between these groups, I hypothesize that the thoracic extensors can be strengthened in a very similar manner as the lumbar extensors. Rather than performing an extension exercise where only lumbar movement is trained, thoracic movement can be trained by itself or in conjunction with lumbar movement as a whole trunk extension exercise, which would be the most time-efficient.

One trend evident throughout the research on lumbar extension training is that the degree to which strength increases occur can meaningfully vary across different joint angles (12, 32, 33, 34, 61, 75, 88). After a 12-week training intervention, isometric lumbar extension torque typically improves for each joint angle tested but to the greatest extent in the most extended lumbar positions, where increases as high as 92-130% have occurred (12, 32, 34). Regardless of whether someone performs a lumbar extension exercise intervention, the least extension torque can be generated in the most extended angle relative to more flexed positions, but the relative difference in strength between these positions decreases following training (12, 31, 32, 33, 34, 56, 61, 88). Some data also indicates that the difference in lumbar extension strength between full extension and flexion is smaller for females compared to males (31).

When reading the lumbar extension exercise studies published since 1989 (which encompasses nearly all of the research on this topic), I couldn’t help but notice the similarity in how they were designed and how one of two researchers helped conduct almost all of them. Every study I found that assessed isolated lumbar extension strength before and after a training intervention selected the MedX lumbar extension machine to measure isometric lumbar extension torque at the same seven different joint angles (12, 22, 23, 32, 33, 34, 37, 61, 62, 75, 88, 101, 108). Furthermore, the overwhelming majority of these studies included a training group using the same MedX machine to perform lumbar extension exercise through the same ROM (i.e., 72° ROM starting in full extension) (12, 22, 23, 32, 33, 34, 61, 75, 88, 101). To my knowledge, significant increases in lumbar spinal extension strength, as tested on the MedX machine, have only ever been measured to occur in participants without low back pain after using the MedX for a training intervention (12, 22, 23, 32, 33, 34, 61, 75, 88, 101, 105). In contrast, Moon et al (2013) did find participants with non-specific low back pain to significantly increase lumbar spinal extension strength on the MedX machine after completing an eight-week bodyweight core training intervention comprised of 14 dynamic exercises or 16 isometric exercises (74). No such increase has occurred in asymptomatic participants after dynamic trunk extension exercise interventions performed on Roman chairs or other machines, despite the high EMG activation of the lumbar extensor muscles during these exercises (34, 60, 62, 86, 108). 

With its ability to limit extraneous movements, the MedX machine does seem to be rather well designed for isolating the lumbar extensors both for testing and training purposes. While using this device or a similar piece of equipment may be the most reliable method of measuring lumbar extension strength, I wonder how much the positive results of these interventions were influenced by using the same machine for both training and testing. One of these studies conducted by Graves et al (1994) assessed how the decision to train with the MedX lumbar extension machine or one of two different lumbar extension machines may affect changes in isometric lumbar extension strength as tested on the MedX machine (34). Notably, sedentary individuals were recruited as subjects, and the two other machines, which were made by Nautilus and Cybex and analyzed together as one group, lacked the restraint that the MedX machine used to restrict any pelvic motion.  

After 12 weeks of training, where subjects performed one set of 8-12 reps until volitional fatigue once per week with the assigned machine, the MedX group significantly increased isometric lumbar extension torque at all seven angles tested, while the other training group experienced no change at any angle. In light of these findings, the authors concluded that “pelvic stabilization is required to isolate and strengthen the lumbar extensor muscles.” Except for the rep tempo (two-second concentric phase and four-second eccentric phase) and ROM (72° for the MedX group and 90° for the other group), the authors did not state how the subjects were actually instructed to perform the exercises. As Graves et al mentioned in the article, the group training without the pelvic restraint could have relied on the larger gluteus maximus and biarticular hamstrings to perform the exercise. A similar compensation may have also occurred in the Roman chair trunk extension interventions that did not improve isolated lumbar extension strength for untrained participants (62, 108). While untrained individuals may naturally use such a movement strategy, this does not mean that specialized equipment that forcibly restricts pelvic motion is required to strengthen the lumbar extensors. If subjects in the unrestrained group were expressly instructed and consistently coached to perform the exercise exclusively as a lumbar extension movement while keeping the pelvis in a fixed position, they may have been able to train their lumbar extensor muscles more effectively. When using a machine that is designed to prevent any possible pelvic motion, no such instruction or coaching is required to target the lumbar extensors because lumbar extension is the only movement that someone could perform to move the weight. Training experience, or lack thereof, could also be a relevant variable. Even with proper instruction and coaching, sedentary individuals can struggle to perform a particular exercise proficiently without compensatory movement. Moving a load exclusively via dynamic lumbar extension when the pelvis is unrestrained requires a degree of coordination and kinesthetic awareness that can take time to develop. 

Several years after Graves et al’s study was published, Mayer et al (2002) conducted a 12-week dynamic lumbar extension exercise intervention where two groups used the MedX machine, but one group trained without the standard pelvic restraint that was utilized by the other group (61).  When tested on the MedX with pelvic restraint, both groups significantly increased isometric lumbar extension strength at all seven joint angles to a similar degree, indicating that pelvic restraint is not necessary to train the lumbar extensor muscles effectively. While this study still used the MedX machine for both groups, I see no reason why lifters cannot employ other dynamic trunk extension exercises with an appropriate technique to enhance trunk extension strength. Compared to the average person who does not perform resistance training, lifters will typically have better body control during an exercise, but minimizing hip extensor contribution during a dynamic trunk extension exercise may still pose a challenge. If you have spent years intentionally minimizing lumbar spine movement while moving at the hip joint when performing exercises that load the trunk extensors, utilizing the opposite strategy during an exercise may feel awkward for a while. Furthermore, some lifters may be psychologically inclined to perform each exercise in a manner that allows them to move the most weight from point A to point B. While this mindset has its advantages in certain situations, such as a powerlifting meet, applying it to smaller accessory exercises can be counterproductive to long-term progression if it impairs the hypertrophic stimulus provided to the target muscles. Performing a trunk extension exercise with minimal hip extensor contribution will require that you use relatively light loads, so I recommend to not let your ego get the best of you and negatively influence your technique in an effort to move more weight.   

Trunk Lateral Flexion Exercise Value 

Training a specific movement can certainly be an effective means of enhancing strength during that movement, but carryover can also exist between different movements. When a particular muscle can meaningfully contribute to the generation of torque for two different joint movements, exclusively training one movement can transfer to enhanced strength during the other untrained movement. Such an effect has been reported to occur by Bourne at al (2017), who conducted resistance training interventions comprised of knee flexion or hip extension exercise (10). Both groups significantly increased knee flexion and hip extension strength, despite exclusively training only one of the movements. These results are congruent with what would be predicted based on the known determinants of muscular force production. While coordination adaptations are specific to the movement that is trained, increased muscle size and motor unit recruitment can transfer to enhanced force production during any movement where the trained muscle functions as a prime mover.  

With the exception of the transversospinalis, all of the core muscles that function to flex or extend the trunk in the sagittal plane while acting bilaterally contribute to laterally flexing the trunk in the frontal plane when acting unilaterally. Similarly, all of the core muscles that function as trunk rotators also function as trunk lateral flexors. Consequently, trunk lateral flexion exercise can engage nearly all of the core muscles simultaneously and may transfer to enhancing strength in every trunk movement. This is not to say that exclusively using trunk lateral flexion exercise would be as effective for developing the core musculature as a combination of core exercises that train all four trunk functions. However, if you wish to train the greatest number of core muscles with a single exercise, a trunk lateral flexion exercise would be quite effective.

With respect to improving squat and deadlift performance, I expect that exclusively adding a direct trunk extension exercise into your program would have greater carryover than exclusively adding a trunk lateral flexion exercise. However, I do suspect that training both types of exercise to some degree may be somewhat more advantageous in the long term than only utilizing trunk extension exercise. In the absence of research, I speculate that muscles that can meaningfully contribute to two different functions, such as the erector spinae, may experience mildly greater long-term development if both functions are trained relative to if only one function is trained.    

Similarly, if a trained individual wishes to maximize hamstring hypertrophy, I along with many other coaches would advocate that the lifter utilizes both knee flexion and hip extension exercises throughout a macrocycle. While variation can be achieved by cycling through different leg curl variations, completely omitting any hip hinge variation such as Romanian deadlifts in a long-term program may be suboptimal for hamstring development. Both knee flexion and hip extension exercises can train all three biarticular hamstring muscles, however, training interventions that measured changes in the size of these muscles suggest that they do not respond in the same manner to these different movements (10, 55). Knee flexion exercise may be the most effective selection to train the semitendinosus, while hip extension exercise with an extended knee position may target the biceps femoris long head and semimembranosus more effectively. Consequently, it is prudent for a trained lifter to include both types of exercise throughout a training macrocycle to facilitate total hamstring development. It has yet to be tested, but I would hypothesize that some of the posterior core muscles (or even certain subregions within a particular muscle) may also be trained more effectively through a combination of both trunk lateral flexion and trunk extension exercise rather than only trunk extension exercise. As previously mentioned, the posterior core musculature encompasses several distinct muscles that are comprised of their own individual regions, so I would not expect that they would all be optimally developed by a single type of trunk exercise. 

For someone who has both strength and physique goals, a trunk lateral flexion exercise can be rather efficient because it can help develop the abdominals and obliques while strengthening the intrinsic back muscles. At a rather high body-fat percentage, the size of the abdominals and obliques will have no discernible effect on someone’s midsection appearance if they are thickly shielded from the world by adipose tissue. However, as the size of the anterior core musculature increases, the body fat percentage required for them to be visible increases in turn. After losing weight following his dominant strongman career, Eddie Hall exemplifies this concept by displaying defined abdominals with a level of body fat that is not particularly low by any metric that applies to mere mortals with normal amounts of muscle. 

Multi-Planar Core Strength           

While trunk extension strength is generally the most impactful type of core strength for recreational lifters and strength athletes, other types of core strength can still directly affect performance for some individuals. Aasa at al’s previously mentioned study, which measured lumbar spine ROM during squats and deadlifts, reported similar amounts of total lateral flexion and rotation motion to occur during these lifts, ranging from 6.2-7.7° (1). The sagittal plane ROM that occurred while squatting and deadlifting was certainly greater, but some frontal and transverse motion may still occur with submaximal loads, so having trunk lateral flexion and rotation strength may help keep this motion from becoming excessive. Trunk lateral flexion strength may be particularly useful while squatting if the barbell begins tilting down to the right or left during a rep. The bar tilt may ultimately stem from a different technique error such as a hip shift or poor starting position, so increasing trunk lateral flexion strength may not address the root issue, but it can help a lifter complete a squat if a noticeable tilt unintentionally occurs.Generally speaking, core strength in multiple planes can become more important as the instability and awkwardness of an exercise increase. As I previously discussed on Stronger By Science, walking out a heavy squat requires that the lifter has a sufficiently high degree of hip abduction strength to stabilize the pelvis in the frontal plane during the moment that one foot leaves the ground when stepping back. However, the frontal plane pelvic stability demands that are present while walking with a very heavy weight may require additional contributions from trunk lateral flexor muscles. McGill et al (2009) analyzed competitive strongman athletes as they performed various strongman events, including a 486lb (220kg) super yoke walk, which is functionally similar to a heavy squat walk out (68). To walk with this load, the athletes required greater hip abduction torque than their hip abductor muscles could maximally generate by themselves. Consequently, they used their quadratus lumborum and obliques, which attach to the pelvis, to provide the additional frontal plane pelvic stability that was needed for the movement. Therefore, strengthening these core muscles may concurrently enhance trunk and hip stabilization, both of which may transfer to a smoother walk out for a heavy squat.     

Programming Recommendations

With all other training variables equated, I doubt that any type of exercise is superior to a dynamic full ROM trunk extension exercise for increasing the size of the intrinsic back muscles. Similarly, full ROM trunk exercises are likely the most effective means of increasing the size of the abdominals, obliques, and quadratus lumborum. Nonetheless, I would not necessarily recommend the inclusion of these exercises for every lifter who prioritizes the development of these muscles to pursue strength and/or physique goals. A wide degree of variation exists among individuals with regard to spinal structure, training experience, prior injury history, and lifestyle factors that may all affect how someone may respond to a dynamic trunk exercise. Consequently, a method of training that one lifter uses to great success may provoke pain for a different person employing the same technique. Regardless of programming, some people with prior injuries can experience low back pain during particular lumbar spine motions even with a very conservative approach to training.

To be clear, this article is not designed to tell any particular individual how or how not to train, and I am not providing any medical advice. Rather, I wish to communicate the information that is available on this matter, so that you can be better informed to make your own personal choices after assessing both the pros and the cons they may bring. With all of this said, not all dynamic trunk exercises are created equal with regard to the reward to risk ratios they exhibit. If you make the personal decision to include these types of exercises in your program and wish to reduce the likelihood that the load tolerance of certain lumbar tissues are exceeded, adhering to a few recommendations may be prudent. Ultimately, I cannot claim that any of these recommendations will definitively reduce the risk of injury for any particular person, but you can assess the reasoning behind them and decide if they merit inclusion in your program. While trunk extension exercises are the focus of these programming recommendations, lifters can apply the same principles for flexion, lateral flexion, and rotation exercises.     

The first guideline is to select an exercise variation that minimizes spinal compressive loading. On one end of the spectrum, you can train the posterior core muscles dynamically at long lengths with a cat back deadlift that starts in full trunk flexion while initiating the pull and finishes in full extension during the lockout. Alternatively, you can utilize an exercise that exclusively trains trunk movement while the knee and hip joints remain in fixed positions. Examples of this include a trunk extension exercise performed on a 45° Roman chair, glute ham developer, or a machine that is specifically designed to do so (i.e., back extension machine). These exercises can be performed in a manner that does not use the powerful hip or knee extensors as prime movers. Consequently, they will require substantially lighter loads to train the posterior core muscles compared to an exercise that does, therefore limiting lumbar compressive forces.   

When minimal equipment is available, you can also use the Jefferson curl to train trunk extension dynamically with free weights. The standard version of this exercise involves hip extension as well, but you can modify it to a variation with a fixed hip position to reduce how much external load is required to target the posterior core muscles. To begin this variation, you can hinge at your hips until reaching the bottom position of a Romanian deadlift with fully stretched hamstrings. Once in this position, you can flex your trunk during the eccentric phase and extend it during the concentric phase while keeping the hips in a statically flexed position. Doing so will require a very low absolute load to train the posterior core muscles dynamically. If you have access to a slant board, it can also be used for the modified Jefferson curl to help deemphasize the hip extensor muscles.

The second guideline is to select a rep range and tempo that further minimizes how much external load will be required to stimulate the desired adaptations. Unsurprisingly, spinal forces are greater when the load or speed used for an exercise is increased (50, 110). While intentionally lifting with an extremely slow velocity (e.g., 10 second concentric phase) is an exception, using a fairly slow and controlled tempo can be just as effective as lifting rapidly for inducing hypertrophy if sets are performed until the same proximity to failure (96, 97). High-load training is typically the most effective way to induce neural adaptations that transfer to improved maximal strength, but a wide range of absolute intensities (i.e., percentages of 1-RM) can be used to maximize muscular hypertrophy (53). If an equal number of sets are performed until volitional failure, a very similar degree of muscle growth may be produced by 6-rep sets, 30-rep sets, or anywhere in between, which may roughly correspond to 50-85% of a 1RM for many lifters (95).  Certainly a minimum intensity threshold exists for a set to induce a robust hypertrophic response, but even lifting loads as light as 30% of 1RM has yielded similar growth as lifting with 80% of 1RM, so a wide range can be utilized (71). Consequently, a dynamic trunk exercise with light loads that enable 20-30 reps per set will likely stimulate core muscle hypertrophy just as effectively as a high-load exercise but with lower peak forces transmitted through the lumbar spine. While Fisher et al (2018) did not examine any changes in muscle size, they did assess how different intensities may affect changes in lumbar extension strength after a lumbar extension training intervention with the MedX machine. Fisher et al assigned recreationally active subjects to a high-load (80% of maximum voluntary isometric contraction) or low-load (50% of maximum voluntary isometric contraction) exercise group for six weeks of lumbar extension training (23). Each group performed a single set until failure once per week, where the high- and low-load groups respectively averaged 8 and 26 reps per set. After the intervention, both groups significantly increased lumbar extension strength to a similar degree, indicating that high or loads can both be effectively utilized. 

The third guideline is to avoid the extreme terminal ranges of motion during a dynamic trunk exercise. While loading a muscle in its stretched position can yield greater hypertrophy than loading it in a moderate or shortened position, no evidence has yet to emerge indicating that training a muscle at its greatest length is superior to training a muscle at 90% of the peak length it could reach. Perhaps a slightly greater magnitude of stretch-mediated hypertrophy may be induced per set by the former approach, but it may present disproportionately greater risks for some individuals. As previously discussed, repeatedly loading the lumbar spine in extreme flexion may be more likely to stress certain structures excessively compared to if a modestly lower angle of peak flexion is reached. Similarly, loading extreme extension, lateral flexion, or rotation may pose their own risks relative to if close to a full ROM is used for a dynamic trunk exercise but extreme end range is avoided. For instance, the lumbar spine’s facet joints can be particularly stressed in fully laterally flexed, rotated, or extended positions (40,58,94).  

The fourth guideline is to abstain from dynamic trunk exercise soon after waking. Generally, I consider the best time to train to be whenever is most convenient and enjoyable for a particular person within the context of his/schedule and individual preferences to bolster adherence. That may entail training 30 minutes after rising in the early morning to complete a session prior to work for some people or training in the late evening after a day’s work for others. Throughout the day, some aspects of athletic performance will often vary. For instance, jumping ability and anaerobic power is typically greater when measured during the afternoon or evening than the morning (44). While these performance fluctuations may result in testing being more reliable when consistently performed at a similar time, I doubt that they will have much of an impact on long-term development. However, meaningful changes occur within the spine’s intervertebral discs throughout the day.  

When the body remains in a lying position for a prolonged period of time, compressive loading on the spine is essentially absent, facilitating the movement of fluid into intervertebral discs, which increase in height as they swell (57). As the intervertebral discs experience compressive loading in an upright posture throughout the day, some of the fluid is driven out of these discs (28). This fluid shift reduces disc height and is the primary reason why people are on average 1% (on average corresponding to .75” [1.9cm]) taller initially upon waking compared to when they go to bed at night (106). The differences in hydration status result in the discs experiencing greater internal fluid pressure soon after waking relative to later in the day (111, 114). This elevated level of pressure restricts available lumbar intervertebral joint ROM and subjects the discs and spinal ligaments to a greater magnitude of loading for a given degree of motion (3). Consequently, dynamically loading the lumbar spine soon after waking will stress these structures to a greater degree compared to if the same movements were performed later in the day. The most pronounced effect will be present within the first hour after waking, as indicated by the finding that 54% of the reduction in height that occurs throughout the day happens during this period (106). With further passage of time, the rate of change declines, so the initial window merits the greatest consideration with respect to when dynamic trunk exercises may pose an elevated risk for some individuals (106).  

The fifth guideline is to implement a gradual progression if you incorporate dynamic trunk exercise into your training program and recognize that this type of movement may not be the best choice for you as an individual. This is the same principle discussed in the programming recommendations for reverse Nordic curls in a previous Stronger By Science article. For nearly any physical activity that is strenuous enough to induce a meaningful physical adaptation, a sudden spike in volume and/or intensity may result in an elevated risk of injury relative to if stressors were progressed at a less rapid rate. Gradual progression can help facilitate proper recovery and adaptation to resistance training, just as it can for running, jumping, and a multitude of different athletic maneuvers. When adding a new movement into your program, I recommend beginning with low volumes rather than immediately starting with high volumes for an exercise to which you are unaccustomed. If you respond well to the new movement and suspect that higher volumes will be beneficial in the overall context of your program, you can always incrementally increase the volume over time. 

With respect to lumbar extension exercise performed on the MedX machine, Carpenter et al (1991) and Steele et al (2015) investigated how different training volumes can affect strength adaptations, and their findings indicate that very low volumes can be rather effective. Carpenter et al assigned untrained subjects to one of four groups that performed a single lumbar extension exercise set of 8-12 reps to failure either once every other week, once per week, twice per week, or three times per week (12). After 20 weeks of training, each training group increased isometric lumbar extension at all tested joint angles, and the magnitude of increase did not significantly differ among the groups. While these results indicate that a very low volume and frequency can be sufficient to induce lumbar extension strength gains in untrained individuals, lifters with more experience may respond differently.    

Steele et al assessed this matter by recruiting participants who had been resistance training for at least six months including two or more sessions per week and “exercises designed to condition the lumbar extensors” (101). The lifters were assigned to a control group or one of two training groups that performed one or three sets of lumbar extension exercise for 8-12 reps to failure once per week for six weeks while not training “any other lumbar conditioning exercises.” Following the intervention, the control group exhibited a significant 8.9% decrease in lumbar extension strength, while the one-set and three-set groups respectively experienced significant 8.3% and 10.7% increases. The authors of this study interpreted these findings to indicate that “set volume does not impact upon strength changes in trained people” with respect to isolated lumbar extension exercise. While the findings do not clearly demonstrate the superiority of higher volumes, I question this conclusion based on the differences between the training groups. The three-set group had greater training experience by an average of two years (five- vs three-year means) and had 18.9% greater baseline lumbar extension strength, which may result in further increases in lumbar extension strength being more difficult to achieve. Consequently, performing multiple weekly sets of lumbar extension exercise may still be more beneficial than only one weekly set for trained individuals, although diminishing returns will be present and a single set per week can still yield a notable effect.

Generally, I don’t find the control group in training studies to be worth specifically discussing primarily because untrained subjects are assessed in most of these studies. This is not to say that a control group comprised of untrained individuals provides no value because it helps determine if training groups improved test results as a result of becoming more experienced with the test. However, Steele et al’s control group of people with an average of three years of resistance training experience provided further insight because their lumbar extension strength significantly decreased over six weeks when they refrained from “lumbar conditioning exercises.” When compared to this reduction in strength, the significant strength increases experienced by the training groups are all the more impressive, reinforcing how even very low volumes of dynamic lumbar extension exercise can have a meaningful effect for experienced individuals.

When to Program Different Core Exercises

With a broad topic such as core training and the large degree of variability that exists among individuals, I cannot make sweeping programming recommendations that apply to everyone.  Ultimately, I intend to communicate the available data and my interpretation of them in a manner that better equips any particular person to make informed decisions for his/her program.  Nonetheless, we can briefly cover some examples of how these ideas can be put into practice to work toward achieving certain goals.

If you wish to prime your core muscles’ spinal stabilization ability prior to lifts like squats and/or deadlifts, isometric core training is an ideal choice. As Greg Nuckols and Dean Somerset previously discussed, isometric exercises such as plank variations, can be valuable tools to improve your neuromuscular trunk control when squatting, particularly when combined with a focus on proper bracing. A 15-minute bout of isometric core training comprised of the side plank, bird dog, and modified curl up has been assessed to increase trunk stiffness acutely after it was performed, underscoring the value of this training when incorporated into a session’s active warmup (51). While it has yet to be tested, I see no reason why other isometric core exercises cannot be used to induce a similar effect. With consistent application, isometric core training can also yield a longer lasting resting core stiffness as indicated by a six-week training intervention that included a variety of isometric core exercises (52). Beyond transiently increasing trunk stiffness, isometric core exercises can also warm up muscles throughout the body, including those that act the hip and shoulder joints. Exercises such as side planks, bird dogs, Pallof presses, suitcase carries, waiter walks, and ab wheel rollouts engage many muscles simultaneously, so they are efficient options to reap the benefits of core training while preparing the rest of the body for subsequent lifts. 

An additional benefit of using specifically isometric core exercises to start your session is that isometric actions generate less muscle damage than eccentric actions (45, 102). With the greater muscle damage that it can produce, a dynamic exercise that includes an eccentric phase may be more fatiguing prior to your session’s primary exercises. This will mainly be relevant when selecting exercises that train the intrinsic back muscles during sessions that include squat and deadlift variations. If you prioritize strength performance on these lifts, you’ll want to implement core exercises in a manner that minimizes any interference. Consequently, I generally do not recommend performing challenging sets of dynamic trunk extension and lateral flexion exercises shortly before your squats and deadlifts.   

To facilitate sufficient recovery between training sessions, I also typically do not recommend programming dynamic trunk extension and lateral flexion exercises the day before sessions that include challenging sets of deadlift and squat variations. Essentially, I usually program these core exercises in a similar manner as accessory exercises used to induce quad and gluteus maximus hypertrophy with respect to when they are performed. For instance, if you squat and/or deadlift on Mondays and Thursdays, leg pressing and lunging on Sundays and Wednesdays can easily fatigue you for your squat and deadlift sessions. Given the importance of the intrinsic back muscles to squats and deadlifts, dynamic trunk extension and lateral flexion exercises can cause similar interference if recovery time is inadequate. To maximize recovery time, I generally program these exercises on the same days where I train squats and deadlifts after the primary lifts are completed, which is the same approach I use with knee extension and hip extension accessory exercises. 

When you perform dynamic trunk flexion and rotation exercises in your program is not particularly consequential because oblique and abdominal force production rarely limit performance in any lifts besides specific core exercises used to target these muscles. If you wish to develop your abdominals and obliques, you can include these exercises into an upper body session, lower body session, or even an active recovery day, and one option wouldn’t necessarily be better than the other. Personally, I believe that the best time to perform these exercises is whenever is most convenient to you, which is the same view I have with scapular protraction, hip abduction, and hip flexion exercises.

Conclusion

In the broader sense of the word, a core is an essential component of something, and this indispensable nature certainly applies to the muscular core’s role in resistance training. Whether you are a competitive strength athlete or a recreational lifter, your core muscles serve a pivotal function in numerous movements. Targeting these muscles through specific accessory exercises can transfer to enhanced performance in a variety of lifts such as squats and deadlifts, so direct core training can be a valuable component of nearly any program for lifters who wish to increase their full body strength.

Image Sources

The three core muscle anatomy images were published by “OpenStax,” are licensed as a Creative Commons work, and can be found here.

The lumbar and thoracic spine anatomy images were published by “BodyParts3D, © The Database Center for Life Science,” are licensed as Creative Commons works, and can be found at here.

The spinal curvature image was published by “Injurymap,” is licensed as Creative Commons works, and can be found here.

The spinal ligament anatomy image was published by “sportEX journals,” is licensed as Creative Commons works, and can be found here.

The Eddie Hall transformation image was published by his Instagram account and can be found here.

The Brian Shaw yoke walk image was published by Rogue Fitness’s Twitter account and can be found here.

Full source list is available on desktop versions of this article.