I just diagnosed a similar problem on my friend's 1998 Honda Accord EX with a four-cylinder engine, except it was the seat back tilt motor. The seat back has been stuck, fortunately in a comfortable near-upright position, but when the switch is pushed, it sometimes moves for a fraction of a second, then stops. After removing the seat from the car (4 bolts and an electrical connector on the bottom of the seat), I turned it upside down to see what was going on. Everything looked fine, and the motor connectors were all in place and secure. I pulled the connectors on the motors and applied 12 VDC from an automotive battery charger to the terminals of each motor. Three of the four motors worked normally, traveling from one extreme position to the other, and back again when I applied power in opposite polarity. The seat back recliner motor, however, was silent. The connector receptacles on the motors have handy test points, so one can easily apply power with probes and measure resistance without having to fish around to touch the recessed connector pins. I also noticed that when I touched the test points with the probes, the functioning motors produced hefty sparks, but the dead motor had a very weak spark. A digital ohmmeter showed the working motors had a resistance of about 2 ohms, while the dead motor read 3.9 ohms. Moreover, when I applied power to the dead motor and measured the resistance immediately thereafter, it was around 10 ohms, and then over 10-30 seconds the resistance decreased to around 3.9 ohms again. Curious, I disassembled the motor in place, removing the end cap that contains the brushes and electrical connector, the rotor and the stator housing with two permanent magnets. I found a copious amount of grease on the commutator and the brushes. Thinking that was the problem, I cleaned everything up with lacquer thinner, applied a sparing amount of synthetic bearing grease to the bronze bushings, carefully assembled everything, put the seat back in the car and tried it. No dice. The seat back motor was still non-functional. After a few more cycles of disassembling the motor, reassembling it and puzzling over its silence, I realized that there was a gadget in series with one of the brushes; it looked like a couple of rectangular copper plates, perhaps a quarter inch by half inch with a thin, gray film between them. A bare braided wire was welded onto one of the connector terminals and one of the copper plates, while another braided wire was welded on the second copper plate and one of the motor brush terminals. Measuring the resistance of the rotor, I found each of the windings read around 1.0 ohm, plus or minus 0.1 ohm. Each of the brushes was worth about 0.4 to 0.5 ohms. The rotor winding plus two brushes adds up to about 2 ohms, the same as the total resistance of one of the known good motors. So, the copper gadget was reading around 2 ohms on its own, and more, depending on whether power had recently been applied. I came to the conclusion that the device is a positive temperature coefficient (PTC) thermistor, a resistor whose resistance varies by the amount of current passing through it. That's how Honda engineers figured out how to build a cheap-ass electromechanical actuator without the need for limit switches and the associated wiring and relays. When the user depresses one of the seat motor switches, it runs the motor to its mechanical limit. When the limit is reached, the motor stalls and the rotor current goes up sharply, causing the thermistor to heat up, limiting the motor current and effectively shutting the motor off until the user releases the switch and lets the motor rest for at least a few seconds until the PTC thermistor cools off. In the case of the seat back motor, however, the PTC thermistor had failed and had too high a resistance without current, and when current was applied, since it had a resistance similar to the motor winding, it would heat up in a fraction of a second, usually before the motor even showed any signs of life, its resistance would zoom into the range of tens of ohms, and there wouldn't be enough current to so much as budge the motor.
My temporary fix was to solder a piece of bare AWG 14 copper wire across the thermistor and reassemble the motor. The seat back works smoothly now as it did before, but one must be mindful to never run it to its extreme limits, lest one burn out the windings or blow the fuse. The eventual solution is to find a motorized seat from a similar-vintage Honda Accord at a salvage yard and pull out one of the motors to repair the seat. I understand that the only way one can get one of these seats repaired at a Honda dealership is have the entire seat pan assembly replaced at a cost of about $1000. If one can salvage a motor, one that reads 1.8 to 2.1 ohms cold, all that is necessary is pull off the end cap with the brushes, the stator housing with the permanent magnets, and the rotor, swapping the good end cap assembly with the bad one. If the gears and screw drives are all intact and working, that saves a heck of lot of difficult disassembly and reassembly work that would involve pressing out roll pins and knurled pins in tight quarters, then pressing them back in.
NOTE: The connectors MUST be removed from the motors to read their resistance or to apply external test power, as the control switches on the side of the seat short the motor connectors when they are in the inactive position. I suspect Honda engineers did this to provide dynamic braking so the motors stop instantly, instead of coasting and overshooting when the control switches are released.