Spines and receptors are proxies, not voltmeters
A dendritic spine is the small protrusion on a neuron where most excitatory synapses sit, so spine density is a rough count of how many synaptic contact points exist. Receptor composition is a different kind of proxy: it says how many AMPA receptors, the molecules that translate glutamate release into electrical current, are parked at the synapse. Neither measure is a direct reading of synaptic strength. Electrical weakening, by contrast, is observed as a smaller postsynaptic response to the same stimulation.
That distinction matters for the question. A drop in spine density or AMPA receptor levels is consistent with weaker synapses, but it does not by itself demonstrate that transmission got weaker, and a stable spine count does not rule out a change in receptor number. The two categories of evidence therefore answer related but not identical questions.
After sleep, spines and AMPA receptors tend to fall
In mice, dendritic spine density decreased during physiological sleep and increased after sleep deprivation, which is the pattern the synaptic homeostasis hypothesis predicts [1]. The same study found that a marker of presynaptic vesicles, SV2A, moved in the opposite direction, falling after sleep deprivation, and that the pre- and postsynaptic changes appeared coordinated across cortical layers [1]. That is a structural signal, not an electrical one, and the study did not record synaptic responses.
A separate rat study looked at AMPA receptors directly and found that sleep lowered AMPAR levels in the hypothalamus as well as in neocortex, with the strongest decreases for receptors containing the GluA1 subunit [5]. By comparing REM sleep, total sleep deprivation and slow-wave sleep, the authors concluded that slow-wave sleep was the driver of this down-regulation, with no additional contribution from REM sleep or circadian rhythm [5]. Again, this is receptor abundance, not measured synaptic current.
The electrical side comes from different experiments
The clearest electrical evidence of synaptic weakening in the supplied work does not involve sleep at all. In rats, a single electrically induced epileptic afterdischarge at the perforant pathway–dentate gyrus synapse produced a long-lasting decrease in evoked postsynaptic potentials, and this functional change was accompanied by shrinkage of mushroom-shaped spines and reduced GluA1-containing AMPA receptors in the postsynaptic density [2]. That study ties electrical weakening, spine shrinkage and receptor loss together within one preparation, but under seizure-like stimulation rather than sleep.
Work on sleep and plasticity describes potentiation and depression of synaptic transmission occurring in parallel during sleep, and frames this as offline synaptic plasticity supporting memory consolidation [3]. That account addresses electrical changes during sleep but does not report spine density or receptor composition, so it cannot be matched point-for-point against the structural findings.
Why the two kinds of evidence have not been reconciled
The gap is largely a matter of experimental design. The sleep studies that measured spines or AMPA receptors did not record electrical synaptic responses [1][5], and the study that measured electrical weakening alongside spine and receptor changes did not use sleep as a condition [2]. The sleep-plasticity review discusses transmission changes without structural measures [3]. No supplied study measured all of these in the same animals under sleep.
There are also signs that the structural story is not uniform. Adolescent mice had higher baseline SV2A and spine density than adults, and their SV2A was more sensitive to sleep deprivation, while adults responded more to physiological sleep [1]. In older adults, post-mortem synaptic density was globally reduced in people with depression, most pronounced in the frontal lobe, and reduced sleep was correlated with depression [4]. That study links sleep and synaptic density only indirectly, through depression, and its authors note that causation is difficult to establish [4]. Taken together, the structural and electrical findings point the same way often enough to be consistent, but the agreement is inferred across studies rather than observed within one.
Sources used in this answer
Sleep differentially modulates synaptic vesicle protein 2A and spine density across cortical regions and development
In mice, dendritic spine density decreased during sleep and increased after sleep deprivation, while the presynaptic marker SV2A fell after sleep deprivation, with developmental differences between adolescent and adult animals.
Single Electrically Induced Epileptic Afterdischarge Triggers Synaptic Weakening, AMPA Receptor Endocytosis, and Shrinkage of Dendritic Spines in the Hippocampus
A single electrically induced epileptic afterdischarge in rats produced long-lasting decreases in evoked postsynaptic potentials alongside shrinkage of mushroom-shaped spines and reduced GluA1-containing AMPA receptors in the hippocampal postsynaptic density.
Offline neuronal activity and synaptic plasticity during sleep and memory consolidation
Potentiation and depression of synaptic transmission are described as occurring in parallel during sleep as offline synaptic plasticity supporting memory consolidation.
Depression in older adults and its associations with sleep and synaptic density
In older adults, depression was associated with reduced sleep and with a global post-mortem reduction in synaptic density, most pronounced in the frontal lobe, though causation could not be established.
Slow-wave sleep drives sleep-dependent renormalization of synaptic AMPA receptor levels in the hypothalamus
Sleep lowered AMPA receptor levels, especially GluA1-containing receptors, in rat hypothalamus and neocortex, with slow-wave sleep identified as the driver rather than REM sleep or circadian rhythm.