The example in your lecture is making a reference to convergence in distribution. Below, I try to go through some of the details of what this means.
A general definition
A sequence of random variables $X_1,X_2,\ldots$ converges in distribution to a limiting random variable $X_\infty$ if their associated distribution functions $F_n(x) = \mathbb P(X_n \leq x)$ converge pointwise to $F_\infty(x) = \mathbb P(X_\infty \leq x)$ for every point $x$ at which $F_\infty$ is continuous.
Note that this statement actually says nothing about the random variables $X_n$ themselves or even the measure space that they live on. It is only making a statement about the behavior of their distribution functions $F_n$. In particular, no reference or appeal to any independence structure is made.
The case at hand
In this particular case, the problem statement itself assumes that each of the $X_n$ have the same distribution function $F_n = F$. This is analogous to a constant sequence of numbers $(y_n)$. Certainly if $y_n = y$ for all $n$ then $y_n \to y$. In fact, we can "map" our convergence-in-distribution problem down to such a situation in the following way.
If we fix an $x$ and consider the sequence of numbers $y_n = F_n(x) = F(x)$, we see that $y_1,y_2,\ldots$ is a constant sequence and so, obviously, converges (to $F(x)$, of course). This holds for any $x$ we choose, and so the functions $F_n$ converge pointwise for every $x$ (in this case) to $F$.
To finish things off, we note that $F(x) = \mathbb P(X_1 \leq x) = \mathbb P(-X_1 \leq x)$ by the symmetry of the normal distribution, so $F$ is also the distribution of $-X_1$. Hence $X_n \to -X_1$ in distribution.
Some equivalent and related statements for this example
To perhaps clarify the meaning of this notion further, consider the following (true!) statements about convergence in distribution, all of which use the same sequence you've defined.
- $X_1, X_2,\ldots$ converges in distribution to $X_1$.
- Fix any $k$. $X_1, X_2,\ldots$ converges in distribution to $X_k$.
- Fix any $k$. $X_1, X_2,\ldots$ converges in distribution to $-X_k$.
- Define $Y_n = (-1)^n X_n$. Then, $Y_n \to X_1$ in distribution.
- Slightly trickier. Let $\epsilon_n$ be random variables such that $\epsilon_n$ is independent of $X_n$ (but, not necessarily of other $\epsilon_k$ or $X_k$) and taking the values $+1$ and $-1$ with probability 1/2, each. Define $Z_n = \epsilon_n X_n$. Then, the sequence $Z_1, Z_2,\ldots$ converges in distribution to $X_1$. This sequence also converges in distribution to $-X_1$ and $\pm X_k$ for any fixed $k$.
Explicit examples incorporating dependence
The easiest way to construct examples in which the $X_i$ are dependent is to use functions of a latent sequence of iid standard normals. The central limit theorem provides a canonical example. Let $Z_1,Z_2,\ldots$ be an iid sequence of standard normal random variables and take
$$X_n = n^{-1/2} \sum_{i=1}^n Z_i \>.$$
Then each $X_n$ is standard normal, so $X_n \to -X_1$ in distribution, but the sequence is obviously dependent.
Xi'an provided another nice (related) example in a comment (now deleted) to this answer. Let $X_n = (1-2 \mathbb I_{(Z_1+\cdots+Z_{n-1}\geq 0})) Z_n$ where $\mathbb I_{(\cdot)}$ denotes the indicator function. The necessary conditions are, again, satisfied.
Other such sequences can be constructed in a similar way.
An aside on the relationship to other modes of convergence
There are three other standard notions of convergence of random variables: almost-sure convergence, convergence in probability and $L_p$ convergence. Each of these are (a) "stronger" than convergence in distribution in the sense that convergence in any of these three implies convergence in distribution and (b) each of these three, in contrast to convergence in distribution, requires that the random variables at least be defined on a common measure space.
To achieve almost-sure convergence, convergence in probability or $L_p$ convergence, we often have to assume some additional structure on the sequence. However, something slightly peculiar happens in the case of a sequence of normally distributed random variables.
An interesting property of sequences of normal random variables
Lemma: Let $X_1,X_2,\ldots$ be a sequence of zero-mean normal random variables defined on the same space with variances $\sigma_n^2$. Then, $X_n \xrightarrow{\,p\,} X_\infty$ in probability if and only if $X_n \xrightarrow{\,L_2\,} X_\infty$, in which case $X_\infty \sim \mathcal N(0,\sigma^2)$ where $\sigma^2 = \lim_{n\to\infty} \sigma_n^2$.
The point of the lemma is three-fold. First, in the case of a sequence of normal random variables, convergence in probability and in $L_2$ are equivalent which is not usually the case. Second, no (in)dependence structure is assumed in order to guarantee this convergence. And, third, the limit is guaranteed to be normally distributed (which is not otherwise obvious!) regardless of the relationship between the variables in the sequence. (This is now discussed in a little more detail in this follow-up question.)
This figure shows the distributions of the means of $n=1$ (blue), $10$ (red), and $100$ (gold) independent and identically distributed (iid) normal distributions (of unit variance and mean $\mu$):
As $n$ increases, the distribution of the mean becomes more "focused" on $\mu$. (The sense of "focusing" is easily quantified: given any fixed open interval $(a,b)$ surrounding $\mu$, the amount of the distribution within $[a,b]$ increases with $n$ and has a limiting value of $1$.)
However, when we standardize these distributions, we rescale each of them to have a mean of $0$ and a unit variance: they are all the same then. This is how we see that although the PDFs of the means themselves are spiking upwards and focusing around $\mu$, nevertheless every one of these distributions is still has a Normal shape, even though they differ individually.
The Central Limit Theorem says that when you start with any distribution--not just a normal distribution--that has a finite variance, and play the same game with means of $n$ iid values as $n$ increases, you see the same thing: the mean distributions focus around the original mean (the Weak Law of Large Numbers), but the standardized mean distributions converge to a standard Normal distribution (the Central Limit Theorem).
Best Answer
Unfortunately, the quoted statement in the question is a muddled version of the one I originally intended. Thank you for catching this. I've also updated the statement in the original question. A counterexample to the former is given at the end of this answer.
Here is the intended statement:
Lemma: Let $X_1,X_2,\ldots$ be a sequence of zero-mean normal random variables defined on the same space with variances $\sigma_n^2$. Then, $X_n \to X_\infty$ in probability if and only if $X_n \xrightarrow{\,L_2\,} X_\infty$, in which case $X_\infty \sim \mathcal N(0,\sigma^2)$ where $\sigma^2 = \lim_{n\to\infty} \sigma_n^2$.
Remark: The main points here are that (a) we can "upgrade" convergence in probability to $L_2$ convergence in the case of sequences of normals, (b) we are guaranteed that the distribution of the limit is normal (which is not otherwise obvious) and (c) we get both of the above without specifying anything about the joint distributions of the elements in the sequence.
Proof (sketch): One direction is easy: Convergence in $L_2$ always implies convergence in probability. To prove to the other direction: If $X_n \to X$ in probability, then $$\varphi_n(t) = \mathbb E e^{it X_n} \to \mathbb E e^{it X_{\infty}}$$ by dominated convergence. But, $\varphi_n(t) = e^{t^2 \sigma_n^2 / 2}$, and $e^{t^2 \sigma_n^2 / 2}$ converges for each $t$ as $n \to \infty$ if and only if $\sigma_n^2 \to \sigma^2$. This is enough to imply that $\sup_n \sigma_n^2 < \infty$, and hence the collection $\{X_n^2\}$ is uniformly integrable. Thus, $X_n \xrightarrow{\,L_2\,} X_\infty$. This further shows the limit $X_\infty$ must be normal since then $\mathbb E e^{it X_\infty} = e^{t^2 \sigma^2 / 2}$, which is the characteristic function of a normally distribution random variable.
Notes
The convergence of the sequence $\sigma_n^2$ and the fact that the limiting distribution $X_\infty$ is normally distributed are part of the conclusion. By the same exact argument, we can also replace $L_2$ convergence with the more general $L_p$ convergence by recognizing that the variance determines the distribution in this case and all moments are finite, so $\{X_n^p\}$ is also uniformly integrable.
From this it is clear that we have the following weaker result on convergence in distribution, which is well-known and given as an exercise in some probability textbooks.
Lemma: Let $X_1,X_2,\ldots$ be a sequence of normal random variables. Then, $X_n \to X_\infty$ in distribution if and only if $\mu_n \to \mu$ and $\sigma_n^2 \to \sigma^2$, in which case $X_\infty \sim \mathcal N(\mu,\sigma^2)$.
A nice application of the second lemma is to consider the marginal distribution of the Riemann integral of Brownian motion, $$ I_t = \int_0^t B_s \, \mathrm{d} s \> . $$ By considering the Riemann sums and using the second lemma, we see that $I_t \sim \mathcal N(0, t^3/3)$.
A counterexample to the quoted statement in the question can be found by considering $X_\infty \sim \mathcal N(0,1)$ and $X_n = (-1)^n X_\infty$. Here, $\sigma_n^2 = 1$ for all $n$ so $\sigma_n^2 \to 1$, but $X_n$ does not converge to $X_\infty$ in probability or $L_2$.