Question about the behavior of Streams.reduce() when the stream is parallel

Question

My question has to do with when one of the parameters in the reduce function of Streams is used, I will need to explain my setup first before to describe the problem more fully.

Suppose I have a Person class with String name and int age as fields, I make a few people in the following static method:

public static List<Person> createPeople() {
    return List.of(new Person("Sara", 20),
                   new Person("Sara", 22),
                   new Person("Bob", 20),
                   new Person("Paula", 32),
                   new Person("Paul", 32),
                   new Person("Jack", 3),
                   new Person("Jack", 72),
                   new Person("Jill", 11));
    }

Now, I want to get a List of the names of people over 30 years of age. I know I can achieve this using the collect function easily, but I wanted to explore one aspect of doing it with the reduce function. I have the following code:

List<String> namesOfOlderThan30 = createPeople().stream()
                                                .filter(person -> person.getAge() > 30)
                                                .map(Person::getName)
                                                .map(String::toUpperCase)
                                                .reduce(new ArrayList<String>(), 
                                                (names, name) -> {
                                                        names.add(name);
                                                        System.out.println(names);
                                                        return names;
                                                 }, 
                                                (names1, names2) -> {
                                                 System.out.println("Before: "+ names1);
                                                 names1.addAll(names2);
                                                 System.out.println("After: "+ names1);
                                                 return names1;
                                           }); 

According to the source, I see that the method signature for reduce when there are three arguments is:

<U> U reduce(U identity,
                 BiFunction<U, ? super T, U> accumulator,
                 BinaryOperator<U> combiner);

When I run this code above, I get the output:

[PAULA]
[PAULA, PAUL]
[PAULA, PAUL, JACK]

Since I coded each of the functions to print, this means that the BinaryOperator combiner parameter was not used at all.

However, when I run the same code using parallelStream() instead of just stream(), I get:

[JACK, PAUL]
Before: [JACK, PAUL, PAULA]
[JACK, PAUL, PAULA]
[JACK, PAUL]
After: [JACK, PAUL, PAULA, JACK, PAUL, PAULA]
Before: [JACK, PAUL, PAULA]
Before: [JACK, PAUL, PAULA, JACK, PAUL, PAULA]
After: [JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA]
Before: [JACK, PAUL, PAULA, JACK, PAUL, PAULA]
After: [JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA]
After: [JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA, JACK, PAUL, PAULA]
...
Exception in thread "main" java.util.ConcurrentModificationException: java.util.ConcurrentModificationException

This makes me think that the third parameter in the reduce method is related to some parallel function of streams, but I do not understand how it is related. Further, when I remove the lines for printing, the error no longer occurs when using parallelStream.

Can someone help explain what is happening here for me in more detail?

Thank you.