Published August 16, 2026·Training Science·10 min read

Spot reduction failed a 1,200-rep-per-session trial. Regional hypertrophy passed MRI. You cannot pick where fat leaves — you can absolutely pick where muscle shows up, and that choice is the entire science of changing your silhouette.

somewhere right now a person is standing sideways in a mirror doing silent geometry. you know the pose. the question under it is the one nobody asks a trainer out loud: is my shape just... assigned? or can i actually change it? the internet's answers are split between "you can't fight genetics babe" and a circuit of side-bends that will allegedly whittle your waist. both are wrong, and both are wrong in ways that have been measured.

the communities that care most about this question have already built the vocabulary. the goal gets called "the Dorito" in some corners of the internet — a torso that reads wide at the top and tapers down. Curve magazine wrote it plainly: "Angles is what changes someone's perception of width." and here is the receipt i genuinely did not expect to be citing: Healthline's training guide quotes a trainer on the core mechanic — "The wider your back and lats are, the smaller your hips are going to appear." the trainer they're quoting is me. Morgan Olson, in someone else's sourced article, saying the exact thesis of this file. so let's earn it with data.

you can't pick where fat leaves

the spot-reduction test has been run at absurd volume, so we never have to argue about it again. researchers had subjects train one leg, three times a week, for 12 weeks — 960 to 1,200 reps per session. that is not a typo. a thousand reps, one leg, per workout. result: the trained leg lost no fat. the body lost fat where it felt like it — upper body down 10.2 percent, trunk down 6.9 percent, whole body down 5.1 percent — everywhere except the region doing the work. the arm version got the MRI treatment: 104 people trained one arm for 12 weeks, and the scanner found no fat difference between the trained and untrained arm.

mechanism, one sentence: fat mobilizes system-wide through hormones responding to an energy deficit, and your DNA — not your exercise selection — decides the withdrawal order. so every "waist-whittling" circuit, every side-bend marathon, every ab-finisher sold as sculpting: they are asking fat cells to read a room they are not in. which sounds like bad news for shape change until you meet the other half of the literature.

you can absolutely pick where muscle shows up

muscle is the opposite of fat: it grows exactly where you send the work, with a precision that still surprises researchers. the field calls it regional hypertrophy. MRI studies by Wakahara found that the regions of a muscle that activate most during an exercise are the regions that grow most after weeks of training it. a 2021 trial found leg extensions grew the entire rectus femoris while Smith-machine squats grew only the central quad — same weeks, same effort, different growth map. and in a trial run entirely on 38 women, preacher curls grew the lower arm region from 2.70 to 2.94 centimeters while incline curls produced no measurable growth at all.

the effect sizes get wild. same leg-curl machine, seated versus lying — the only difference is hip angle, which stretches the hamstrings more — produced +14 versus +9 percent growth, roughly 55 percent more from choosing the other seat. triceps trained overhead instead of at your side grew 1.4 times as much — 1.5x in the long head — despite using lighter weights. read that pair again: the exercise menu is not a style choice. it is a set of coordinates for where your next kilogram of muscle gets delivered. honesty note before we build with this: several of these comparisons measure activation [EMG], and a 2023 trial showed higher activation does not always mean more growth — so where we have direct MRI growth data, we use it first.

the silhouette blueprint, muscle by muscle

so the strategy writes itself: stop trying to subtract from the bottom of the V. build the top of it. the width you see in a frame comes from three trainable places, and each has a best-evidence pick:

that last row deserves a sentence in plain speak, because it cuts both ways and this audience contains both directions. glutes respond hugely to hip thrusts — EMG in 13 women measured 69.5 to 86.8 percent activation versus 29.4 to 45.4 in the squat — but the 2023 MRI trial found both lifts grow the glutes the same. the real difference is the passengers: squats bring the quads and inner thighs along, hip thrusts don't. want more lower-body mass to balance a frame — squat. want glute strength without adding size around the hips — thrust. neither answer is wrong. the point is that it's a dial you get to turn, not a verdict you receive. [that trial is also the receipt for the EMG caveat above: the thrust's higher activation did NOT produce more growth. activation is a map, not a guarantee.]

your hips are bone. your shoulders are negotiable.

one more layer, handled carefully because it deserves care. the shoulder-to-hip ratio is a real perceptual variable — EEG studies show human brains register silhouette proportions pre-consciously, before you've formed a single opinion. that is ALL that research licenses: the silhouette signal is real. nobody owes anyone a ratio. your pelvis is a fixed piece of architecture and it is not a problem to solve — the entire trainable variable sits above it, which is exactly why the communities that think hardest about this converged on the same program from lived experience. QueerDoc's physique guide: "rows, pullups, and pushups to... add width to the back," big quads to balance the frame. TransAware: lateral raises and overhead presses for width. the lived-experience playbook and the MRI playbook are the same playbook. that convergence is what a real answer looks like.

and the timeline honesty, because we owe you numbers over vibes: the delt-growth figures above came from trained lifters over 8 weeks — a few percent per cycle, compounding. this is a seasons project, not a 30-days project [the honest timeline file has the full math]. the silhouette changes in the order the work goes in: strength first, then shape, then strangers noticing you look "different" and not being able to say why. they can't say why because the why is a dumbbell held out level at shoulder height, twice a week, for a year.

the week, written down

everything above, converted into the part you can run today. two upper days, one lower day, inside whatever program you already run — or as the skeleton of a new one:

form notes in one breath: lateral raises stop at shoulder height with a soft elbow — higher is trap territory. pulldowns come to the collarbone, not behind the neck [the comparison data is unambiguous]. rows finish with a squeeze you could pause. and if the gym floor itself is the obstacle, the unwatched workout runs every one of these movements in one corner with dumbbells.

---

quick answers

can you actually change your body shape by lifting?

yes — through muscle, not fat. you cannot choose where fat leaves: a 12-week trial that trained one leg with 960 to 1,200 reps per session found no fat loss in the trained leg while the upper body lost 10.2 percent. but you can choose where muscle grows: MRI studies show exercise selection controls which muscles, and even which regions of a muscle, get bigger. shape change is a construction project, not a subtraction project.

can i make my hips look smaller with training?

hip bones don't move, and no exercise burns fat off one spot. what changes is proportion: building the shoulders, lats, and upper back makes the waist and hips read narrower by comparison. exercise choice also matters lower down — hip thrusts and squats grew glutes equally in a 2023 MRI trial, but the squat also grew the quads and adductors while the hip thrust left them unchanged, so you can pick which muscles come along.

what exercises actually build wider shoulders?

abduction work — lateral raises first. an 8-week trial found lateral raises grew the side delts 3.3 to 4.6 percent, with dumbbells and cables equally effective. EMG comparisons find the lateral raise with a neutral arm position activates the side delt best, while front raises mostly work the front delt and chest — they build very little width. rows, pull-ups, and pulldowns handle the lats, the other half of the taper.

does a wide grip build wider lats?

no. a controlled comparison of pulldown grip widths at 1x and 2x shoulder width found equal lat activation, and the medium grip actually moved slightly more weight — 6RM of 80 kilograms versus 77.3 wide. width comes from making the lats bigger with full-range pulldowns and rows, not from where your hands go on the bar.

keep reading

• where do broad shoulders actually come from. — the deep file on the top of the V
• why don't i look like i lift yet. — the honest timeline this whole project runs on
• what does toned actually mean. — the marketing autopsy behind the subtraction myth
• am i going to hurt my back deadlifting. — the pull that anchors every back program

---

references

educational content, not a prescription — the growth figures come from specific trained populations and your first cycles will write their own numbers. delt rates cited are from trained lifters; beginners typically see faster early progress.

- MoMo