These consolidated cases raise the question of whether there is a seizure, as a matter of law, whenever the police pull up behind a stopped car and turn on their flashing lights. We hold that there is not, but because the trial court may not have appreciated the principles governing the law of stops and seizures in its denial of defendants’ motions to suppress, we remand these cases for redetermination in light of our clarification of the applicable law.
The facts are that the officer noticed a car, with its headlights on and engine off, parked on the side of a street in front óf a vacant lot in a business district at about one in the morning. One of the defendants was leaning into the back seat of the car. The officer thought the car had broken down or that something suspicious was occurring, so he pulled up behind the car and turned on his emergency lights.
The evidence was disputed concerning whether defendants were free to leave. The officer repeatedly testified that they were free to leave at all times. On the other hand, he also testified that they were not free to leave for a few minutes after he turned on his lights, and he was impeached with prior testimony in which he said that persons stopped are not free to leave when his lights are engaged.
The officer approached defendants’ car and looked inside. He saw two open containers of alcohol and a stereo receiver. He was told that the car had run out of gas. He suspected that one defendant was intoxicated. He obtained permission to search the car and check the serial number on the stereo. He asked defendants to pour out the remaining alcohol before defendants left to get gas. A short time later, the officer learned that the stereo had been stolen in a recent burglary. The officer then found defendants and arrested them.
*108Although we have written the above paragraph in sentences indicating a time sequence, the testimony was not at all clear that the time sequence was as set out above. Neither party questioned the officer about exactly what he said and did as he approached the vehicle. Defendants did not testify.
Both the applicable law and the standard of review to be utilized in this case have recently been set forth in State v. Lopez, 109 N.M. 169, 783 P.2d 479 (Ct. App.), cert. quashed, 109 N.M. 131, 782 P.2d 384 (1989). The law is that a person is seized within the meaning of the fourth amendment (and thus the police must justify the seizure by probable cause or reasonable suspicion) when, in view of all the circumstances, the person is accosted and restrained such that a reasonable person would have believed he or she was not free to leave. Id. at 170, 783 P.2d at 480. The standard of review is that if different inferences can be drawn from the facts, the question of whether a person is accosted and restrained in such a way is a factual question subject to the substantial evidence standard. Id. This standard is significantly different from the “independent judgment” standard set forth in People v. Bailey, 176 Cal.App.3d 402, 222 Cal.Rptr. 235 (Ct.App.1985), relied on in the dissent.
Defendants argue that the Lopez standard of review essentially abrogates the de novo review that they contend is required whenever there is not a direct conflict in the testimony. The state argues the evidence in this case as though no question of law whatsoever is involved. We disagree with both parties’ extreme positions and take this opportunity to clarify Lopez’s dual standard of review.
Lopez’s statement of the standard of review is not, as defendants contend, an “irrational” and “artificial” thwarting of the appellate court’s proper role. Rather, it is a simple recognition that trial courts are in a better position than appellate courts to find the facts, and that such fact-finding frequently involves determining which inferences to draw.
For example, in this case, the testimony of the officer involved was internally contradictory as to whether defendants were free to leave. As an initial factual question, it was for the trial court to determine whether, in the officer’s mind, defendants were free to leave or not. Of course, what is in the officer’s mind is not determinative. The issue under Lopez is how a reasonable person in defendants’ circumstances would have felt. Nonetheless, people have nonverbal ways of communicating what is on their minds, and a trial court could find, based on what is on an officer’s mind together with surrounding circumstances, that if the officer believes that the defendants are not free to leave it may be more likely that the defendants would feel that they are not free to leave. The contrary would also be true: if the trial court finds that the officer believed that the defendants were free to leave, it may be more likely that they would feel they are free to leave.
The above discussion shows that factual conflicts are to be determined by the trial court. It further shows that even when the facts do not appear to be in dispute, it is possible that different inferences may be drawn from the facts. When such is the case, it is for the trial court to decide the facts, including the drawing of inferences. See State v. McGhee, 103 N.M. 100, 103, 703 P.2d 877, 880 (1985). Defendants’ complaint that this could result in two disparate decisions on the same facts is essentially a complaint about the nature of appellate review. Yet, it is well established that it is inherent in the nature of review that different trial courts may reach different conclusions, and that does not compel a reversal. State v. Anderson, 107 N.M. 165, 168, 754 P.2d 542, 545 (Ct.App.1988). Defendants’ complaint that different conclusions should not be allowed on the same facts in constitutional cases ignores the fact that there are rarely cases with identical facts.
Nonetheless, it appears to us that the trial court in this case may have misapplied the law in applying it to the facts as found. Actually, we do not know exactly what facts the trial court found. Defendants’ *109suppression motion was denied in an order without stating reasons. Additionally, we do not even know exactly on what theory the trial court denied the motion to suppress. The state had argued that (1) there was not a stop; (2) if there was, it was supported by reasonable suspicion; and (3) even if not, defendants freely consented to the search under permissible circumstances. In cases such as this, in which we do not know the trial court’s rationale, particularly when the evidence supporting the trial court’s decision is as thin as it is here, it is appropriate to remand to the trial court for a redetermination in accordance with the law that we are clarifying today. See, e.g., State v. Tindle, 104 N.M. 195, 200, 718 P.2d 705, 710 (Ct.App.1986). We shall explain why we believe the evidence supporting the trial court’s decision is thin after we discuss the applicable law.
Both parties cite cases involving the use of flashing lights by police officers. To the extent that defendants’ cases involve moving vehicles' that stopped in response to the flashing lights, we believe those cases are distinguishable and of little assistance. To the extent that the state’s cases involve pedestrians unaware that the flashing lights were intended to be signal-ling them, they too are distinguishable and of little assistance.
Two cases, however, deserve more discussion. They are State v. Walp, 65 Or. App. 781, 672 P.2d 374 (1983), and State v. Stroud, 30 Wash.App. 392, 634 P.2d 316 (1981), review denied, 96 Wash.2d 1025 (1982). Both involved factual circumstances similar to those here in that the officers pulled up behind stopped vehicles and activated their emergency lights. Walp was a defendant’s appeal in which the trial court’s ruling that there was no seizure was reversed. Stroud was a state’s appeal in which the trial court’s ruling that there was a seizure was affirmed. While the procedural posture of Walp better supports defendants’ position here, we are not impressed with Walp’s reasoning or result and do not adopt it in New Mexico. While the procedural posture of Stroud does not support defendants’ position as much, we are more impressed with its reasoning and commend it to the trial court on remand here.
Walp involved a woman in a stopped car. An officer thought the woman was having mechanical difficulty and turned on his lights to investigate. Walp was based in part on a statute making it a crime to drive after police lights are activated. However, to the extent that it holds that, as a matter of law, a stop that must be supported by at least reasonable suspicion occurs whenever lights are activated, regardless of the officer’s motive and actions and regardless of facts supporting a belief that the stopped driver is free to leave, we disagree with it.
We can conceive of many situations in which people in stopped cars approached by officers flashing their lights would be free to leave because the officers would be simply communicating with them to ascertain that they are not in trouble. Under such circumstances, depending on the facts, the officers may well activate their emergency lights for reasons of highway safety or so as not to unduly alarm the stopped motorists. We are loathe to create a situation in which officers would be discouraged from acting to help stranded motorists, from acting in the interest of the safety of the travelling public, or from acting in the interest of their own safety.
On the other hand, we find it hard to conceive of a situation where officers activate their emergency lights to investigate a suspicious situation and approach the situation with many accusatory questions in which a reasonable stopped motorist would feel free to leave. We view Stroud as an example of such a case. The Stroud court relied on a statute similar to New Mexico’s resisting arrest statute. The court concluded that the defendants there were seized because they arguably could have been charged under the statute had they left. The appellate court agreed with the trial court that this was a show of authority sufficient to convey to a reasonable person that departure was not a realistic alternative. While we have a similar statute in New Mexico, NMSA 1978, Section 30-22-1(C) (Repl.Pamp.1984), we do not believe *110that it would apply to a driver already stopped when the officer approached. This statute proscribes refusing to bring the vehicle to a stop. It does not apply to stopped vehicles. (Of course, if the stopped driver knows that the officer is trying to effectuate an arrest, then NMSA 1978, Section 30-22-l(B) (Repl.Pamp.1984) would apply if the driver tried to leave.)
As indicated above, the trial court has a difficult and sensitive task on remand. It should focus on the question set forth in Lopez: whether, due to physical restraint or a show of authority, a reasonable person in defendants’ situation would feel free to leave under all of the circumstances of the case. The trial court should consider the officer’s subjective intent only to the extent that it would bear on the beliefs of reasonable people in defendants’ shoes. The trial court should consider the statute, § 30-22-l(C), to the same extent. The trial court should consider the sequence of the officer’s actions and determine how that would bear on the beliefs of reasonable people being confronted in the same manner.
By way of example, we believe that a trial court should ordinarily find a stop that must be justified by reasonable suspicion whenever officers pull up behind a stopped car, activate their lights, and approach the car in an accusatory manner, asking for license and registration and an account of the occupants’ activities. On the other hand, a trial court should ordinarily find no stop whenever officers pull up behind a stopped car, activate their lights, and approach the car in a deferential manner asking first whether the occupants need help.
Finally, we leave to the trial court’s discretion the choice of whether or not to take any additional testimony. While there are many questions on which the evidence could have been more clear, the trial court may wish to find on these questions against the party with the burden of proof instead of taking new evidence. In this case, on the issue of whether there was a stop that rises to the level of a seizure, it appears that defendants bear the burden of proof. See 4 Wayne R. LaFave, Search and Seizure § 11.2(b) & n. 45.3 (2d ed. 1987 & pkt. part 1992) (citing Russell v. State, 717 S.W.2d 7 (Tex.Crim.App.1986) (en banc)). This is consistent with New Mexico law to the effect that defendants have the burden to raise an issue as to their illegal search and seizure claims. See State v. Gardner, 95 N.M. 171, 175, 619 P.2d 847, 851 (Ct.App.1980), cert. denied, October 6, 1980. Once they have done so, the burden shifts to the state to justify the warrantless search. See State v. Mann, 103 N.M. 660, 663, 712 P.2d 6, 9 (Ct.App. 1985), cert. denied, 103 N.M. 740, 713 P.2d 556 (1986).
The state contends that even if a stop amounting to a seizure is found, it was supported by ample reasonable suspicion and defendants consented to the search of the car in any event. We disagree. The degree of suspicion in this case was no more reasonable than that we held-to be insufficient in State v. Galvan, 90 N.M. 129, 560 P.2d 550 (Ct.App.1977). For the consent to be valid, it must be attenuated from seizure. State v. Bedolla, 111 N.M. 448, 456, 806 P.2d 588, 596 (Ct.App.), cert. denied, 111 N.M. 416, 806 P.2d 65 (1991). The facts of this case show no attenuation. See id.
We reverse and remand this case for the trial court to redetermine the issues on the motion to suppress in accordance with the views expressed herein.
IT IS SO ORDERED.
ALARID, C.J., concurs.
APODACA, J., dissents.