Composition what's best between my two approaches

this is a theoretical question about composable OOP in JS.

Suppose I want to reason around objects in terms of their behaviors, which one of the approaches below is best and why?

Approach #1:

var createHero = (fname, lname) => ({ firstName: fname, lastName: lname })

var canWalk = self => {
    self.walk = () => (console.log('Walking...'), self)
}

var withUtilityBelt = self => {
    var powers = []
    self.addPower =  power => (powers.push(power), self)
    self.getPowers = () => powers.slice()
}

var combine = (self, behaviors) => {
    behaviors.forEach(b => b(self))
}


var batman = createHero('Bruce', 'Wayne')

combine(batman, [canWalk, withUtilityBelt])

batman
    .walk()
    .addPower('money')
    .addPower('technology')
    .addPower('intelligence')

console.log(batman.getPowers())

Approach #2:

var createHero = (fname, lname) => ({ firstName: fname, lastName: lname })

var canWalk = self => {
    return {
        walk: () => (console.log('Walking...'), self)
    }
}

var withUtilityBelt = self => {
    var powers = []
    return {
        addPower: power => (powers.push(power), self),
        getPowers: () => powers.slice()
    }
}

var combine = (self, behaviors) => {
    behaviors = behaviors.map(b => b(self))
    return Object.assign(self, ...behaviors)
}


var batman = createHero('Bruce', 'Wayne')

combine(batman, [canWalk, withUtilityBelt])

batman
    .walk()
    .addPower('money')
    .addPower('technology')
    .addPower('intelligence')

console.log(batman.getPowers())

I have the sensation that the latter is better than the former as behaviors are objects themselves, yet I don't know if there's any practical advantage in it.

What do you think? Am I missing something useful from a composition point of view??

Thanks in advance! :D